[ { "path": ".coveragerc", "content": "[run]\nbranch = True\n\n[report]\nskip_covered = True\nskip_empty = True\n# show_missing = True\n\n[html]\nskip_covered = True\nskip_empty = True\n" }, { "path": ".github/FUNDING.yml", "content": "github: derneuere\ncustom: https://www.paypal.com/donate/?hosted_button_id=5JWVM2UR4LM96\n" }, { "path": ".github/ISSUE_TEMPLATE/bug_report.md", "content": "---\nname: Bug report\nabout: Create a report to help us improve\ntitle: ''\nlabels: bug\nassignees: ''\n\n---\n\n# 🛑 Before you create an issue make sure that:\n- Your issue is **strictly related to LibrePhotos** itself. Questions about setting up a reverse proxy belong in what ever reverse proxy you are using. \n- You have read the [documentation](https://docs.librephotos.com) thoroughly.\n- You have searched for a similar issue among all the former issues (even closed ones).\n- You have tried to replicate the issue with a clean install of the project.\n- You have asked for help on our Discord server [LibrePhotos](https://discord.gg/xwRvtSDGWb) if your issue involves general \"how to\" questions\n\n**When Submitting please remove every thing above this line**\n\n\n# 🐛 Bug Report\n\n* [ ] 📁 I've Included a ZIP file containing my librephotos `log` files\n* [ ] ❌ I have looked for similar issues (including closed ones)\n* [ ] 🎬 (If applicable) I've provided pictures or links to videos that clearly demonstrate the issue \n\n## 📝 Description of issue:\n\n\n## 🔁 How can we reproduce it:\n\n\n## Please provide additional information:\n- 💻 Operating system: \n- ⚙ Architecture (x86 or ARM): \n- 🔢 Librephotos version: \n- 📸 Librephotos installation method (Docker, Kubernetes, .deb, etc.): \n * 🐋 If Docker or Kubernets, provide docker-compose image tag:\n- 📁 How is you picture library mounted (Local file system (Type), NFS, SMB, etc.): \n- ☁ If you are virtualizing librephotos, Virtualization platform (Proxmox, Xen, HyperV, etc.): \n" }, { "path": ".github/ISSUE_TEMPLATE/enhancement-request.md", "content": "---\nname: Enhancement request\nabout: Suggest an idea for this project\ntitle: ''\nlabels: enhancement\nassignees: ''\n\n---\n\n**Describe the enhancement you'd like**\nA clear and concise description of what you want to happen.\n\n**Describe why this will benefit the LibrePhotos**\nA clear and concise explanation on why this will make LibrePhotos better.\n\n**Additional context**\nAdd any other context or screenshots about the enhancement request here.\n" }, { "path": ".github/workflows/docker-publish.yml", "content": "on:\n push:\n # Publish `dev` as Docker `latest` image.\n branches:\n - dev\n\njobs:\n # Run tests.\n # See also https://docs.docker.com/docker-hub/builds/automated-testing/\n test:\n runs-on: ubuntu-latest\n\n steps:\n - uses: actions/checkout@08eba0b27e820071cde6df949e0beb9ba4906955 # v4\n - name: Run tests\n run: echo \"to-do\"\n\n # Push image to GitHub Packages.\n # See also https://docs.docker.com/docker-hub/builds/\n push:\n # Ensure test job passes before pushing image.\n needs: test\n\n runs-on: ubuntu-latest\n if: github.event_name == 'push'\n\n steps:\n - name: Repository Dispatch\n uses: peter-evans/repository-dispatch@v3\n with:\n token: ${{ secrets.REPO_ACCESS_TOKEN }}\n repository: librephotos/librephotos-docker\n event-type: backend-commit-event\n" }, { "path": ".github/workflows/pre-commit.yml", "content": "name: Linting (using pre-commit)\n\non: [push, pull_request]\n\njobs:\n pre-commit:\n runs-on: ubuntu-latest\n steps:\n - uses: actions/checkout@v4\n - name: Install dependencies\n run: pip install ruff\n - name: Run pre-commit check\n uses: pre-commit/action@v3.0.1\n" }, { "path": ".gitignore", "content": "# Byte-compiled / optimized / DLL files\n__pycache__/\n*.py[cod]\n*$py.class\n\n# C extensions\n*.so\n\n# Docker\nDockerfile\nentrypoint.sh\n\n# Static assets\nstatic/\n\n# Nuitka\nmanage.build/\nmanage.dist/\n\n# Distribution / packaging\n.Python\nbuild/\ndevelop-eggs/\ndist/\ndownloads/\neggs/\n.eggs/\nlib/\nlib64/\nparts/\nsdist/\nvar/\nwheels/\nshare/python-wheels/\n*.egg-info/\n.installed.cfg\n*.egg\nMANIFEST\n\n# PyInstaller\n# Usually these files are written by a python script from a template\n# before PyInstaller builds the exe, so as to inject date/other infos into it.\n*.manifest\n*.spec\n\n# Installer logs\npip-log.txt\npip-delete-this-directory.txt\n\n# Unit test / coverage reports\nhtmlcov/\n.tox/\n.nox/\n.coverage\n.coverage.*\n.cache\nnosetests.xml\ncoverage.xml\n*.cover\n*.py,cover\n.hypothesis/\n.pytest_cache/\ncover/\ncoco/\nsurvey/\n\n# Translations\n*.mo\n*.pot\n\n# Django stuff:\n*.log\nlocal_settings.py\ndb.sqlite3\ndb.sqlite3-journal\n\n# Flask stuff:\ninstance/\n.webassets-cache\n\n# Scrapy stuff:\n.scrapy\n\n# Sphinx documentation\ndocs/_build/\n\n# PyBuilder\n.pybuilder/\ntarget/\n\n# Jupyter Notebook\n.ipynb_checkpoints\n\n# IPython\nprofile_default/\nipython_config.py\n\n# pyenv\n# For a library or package, you might want to ignore these files since the code is\n# intended to run in multiple environments; otherwise, check them in:\n# .python-version\n\n# pipenv\n# According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.\n# However, in case of collaboration, if having platform-specific dependencies or dependencies\n# having no cross-platform support, pipenv may install dependencies that don't work, or not\n# install all needed dependencies.\n#Pipfile.lock\n\n# PEP 582; used by e.g. github.com/David-OConnor/pyflow\n__pypackages__/\n\n# Celery stuff\ncelerybeat-schedule\ncelerybeat.pid\n\n# SageMath parsed files\n*.sage.py\n\n# Environments\n.env\n.venv\nenv/\nvenv/\nENV/\nenv.bak/\nvenv.bak/\n\n# Spyder project settings\n.spyderproject\n.spyproject\n\n# Rope project settings\n.ropeproject\n\n# visual studio\n.vs/\n.vscode/\n\n# IntelliJ\n.idea/\n\n# mkdocs documentation\n/site\n\n# mypy\n.mypy_cache/\n.dmypy.json\ndmypy.json\n\n# Pyre type checker\n.pyre/\n\n# pytype static type analyzer\n.pytype/\n\n# Cython debug symbols\ncython_debug/\n\n# LibrePhotos\ndensecap/data/models/densecap/densecap-pretrained-vgg16.t7\n*/*.pkl\n*/*/*.pkl\nthumbnails\nmedia\nsamplephotos\nConv2d.patch\nLinear.patch\nSequential.patch\nBatchNorm2d.patch\nAvgPool2d.patch\nReLU.patch\nrun_docker.sh\nlogs/\nplayground\napi/im2txt/data/\napi/im2txt/models/\napi/im2txt/png/\n*.ipynb\napi/im2txt/*.tar.gz\n*.db\nmedia*\nprotected_media\nreport_dynamo_export.sarif\n\n# Vim\n*.swp\n*.swo\n" }, { "path": ".pre-commit-config.yaml", "content": "default_language_version:\n python: python3\n\nrepos:\n - repo: https://github.com/astral-sh/ruff-pre-commit\n rev: v0.9.4 # Check latest version\n hooks:\n - id: ruff\n args: [\"--fix\"] # Fix lint & format\n\n - id: ruff-format # Format code\n" }, { "path": "CLAUDE.md", "content": "# LibrePhotos Backend Agent Guidelines\n\n## Build & Development Commands\n\n**Note:** All commands should be run inside the backend Docker container (`docker exec -it backend bash`).\n\n### Django Management\n- **Run Migrations**: `python manage.py migrate`\n- **Make Migrations**: `python manage.py makemigrations`\n- **Create Superuser**: `python manage.py createsuperuser`\n- **Collect Static**: `python manage.py collectstatic`\n- **Shell**: `python manage.py shell`\n- **Custom Commands**: `python manage.py ` (see `api/management/commands/`)\n\n### Running Services\n- **API Server (Gunicorn)**: Runs automatically in container\n- **Background Jobs (django-q2)**: Runs automatically via `qcluster` command\n- **Image Similarity Service**: Flask app for semantic search\n- **Thumbnail Service**: Separate process for image processing\n\n### Linting & Formatting\n- **Lint**: `ruff check .`\n- **Format**: `ruff format .`\n- **Lint + Fix**: `ruff check --fix .`\n\n### Testing\n- **Run All Tests**: `python manage.py test api.tests`\n- **Run Specific Test**: `python manage.py test api.tests.test_module`\n- **Run with Verbosity**: `python manage.py test api.tests -v 2`\n\n### Debugging\n- **PDB Breakpoint**: Add `import pdb; pdb.set_trace()` in code\n- **Attach to Container**: `docker attach $(docker ps --filter name=backend -q)`\n- **Silk Profiler**: Access `/api/silk` (dev mode only)\n- **Detach**: `Ctrl+P` then `Ctrl+Q`\n\n## Code Style & Conventions\n\n- **Formatting**: Ruff with 88 char line width (configured in `pyproject.toml`)\n- **Imports**: Sorted by isort (via Ruff)\n- **Target Python**: 3.11+\n- **Framework**: Django 5.x with Django REST Framework\n- **Async Jobs**: django-q2 with ORM broker\n- **ML Framework**: PyTorch for machine learning models\n\n## Project Structure\n\n### `api/` - Main Application\n- `models/` - Django ORM models (Photo, Face, Person, Album, etc.)\n- `views/` - API endpoints using Django REST Framework\n- `serializers/` - JSON serialization for models\n- `management/commands/` - CLI commands (`python manage.py `)\n- `migrations/` - Database migrations\n- `tests/` - Test suite\n- `geocode/` - Reverse geocoding functionality\n- `feature/` - Feature extraction utilities\n\n### `service/` - Microservices\n- `clip_embeddings/` - CLIP model for semantic search\n- `face_recognition/` - Face detection and recognition\n- `image_captioning/` - Image captioning (im2txt, BLIP)\n- `thumbnail/` - Thumbnail generation\n- `llm/` - LLM integration for chat features\n- `tags/` - Tag extraction (places365)\n- `exif/` - EXIF metadata extraction\n\n### `image_similarity/` - Similarity Search\n- FAISS-based image retrieval index\n- Flask REST API for similarity queries\n\n### Key Files\n- `manage.py` - Django management script\n- `requirements.txt` - Python dependencies\n- `pyproject.toml` - Ruff/project configuration\n- `librephotos/settings/` - Django settings (base, dev, prod)\n- `librephotos/urls.py` - URL routing\n\n## Environment Variables\n\nKey environment variables (set in Docker or `.env`):\n- `DEBUG` - Enable debug mode (0 or 1)\n- `SECRET_KEY` - Django secret key\n- `DB_*` - Database connection settings\n- `MAPBOX_API_KEY` - For map features\n- `WEB_CONCURRENCY` - Gunicorn worker count\n\n## Common Patterns\n\n### Adding a New API Endpoint\n1. Create/update model in `api/models/`\n2. Create serializer in `api/serializers/`\n3. Create view in `api/views/`\n4. Add URL in `librephotos/urls.py`\n5. Run migrations if model changed\n\n### Adding a New Background Job\n1. Define task function in `api/all_tasks.py` or relevant module\n2. Use `@shared_task` decorator for django-q2\n3. Queue with `async_task()` or schedule in admin\n\n### Adding a New ML Model\n1. Add model loading in `api/ml_models.py`\n2. Create service wrapper in `service//`\n3. Integrate with API views as needed\n\n" }, { "path": "CONTRIBUTING.md", "content": "# Contributing to LibrePhotos\n\nThank you for your interest in contributing to LibrePhotos! This guide will help you get started with the development process.\n\n## Table of Contents\n\n- [Development Environment Setup](#development-environment-setup)\n- [Docker & Docker Compose](#docker--docker-compose)\n- [IDE Recommendations](#ide-recommendations)\n- [Code Quality Standards](#code-quality-standards)\n- [How to Open a Pull Request](#how-to-open-a-pull-request)\n- [Getting Help](#getting-help)\n\n---\n\n## Development Environment Setup\n\n### Prerequisites\n\n- **Git** - for version control\n- **Docker** and **Docker Compose** - for running the development environment\n- **Node.js 18+** and **Yarn** - for frontend development (optional, if developing outside Docker)\n- **Python 3.11+** - for backend development (optional, if developing outside Docker)\n\n### Step 1: Clone the Repositories\n\nCreate a directory for the project and clone all required repositories:\n\n**Linux/macOS:**\n```bash\nexport codedir=~/dev/librephotos\nmkdir -p $codedir\ncd $codedir\n\ngit clone https://github.com/LibrePhotos/librephotos-frontend.git\ngit clone https://github.com/LibrePhotos/librephotos.git\ngit clone https://github.com/LibrePhotos/librephotos-docker.git\n```\n\n**Windows (PowerShell):**\n```powershell\n$Env:codedir = \"$HOME\\dev\\librephotos\"\nNew-Item -ItemType Directory -Force -Path $Env:codedir\nSet-Location $Env:codedir\n\ngit clone https://github.com/LibrePhotos/librephotos-frontend.git\ngit clone https://github.com/LibrePhotos/librephotos.git\ngit clone https://github.com/LibrePhotos/librephotos-docker.git\n```\n\n### Step 2: Configure Environment\n\nNavigate to the `librephotos-docker` directory and create your `.env` file:\n\n```bash\ncd librephotos-docker\ncp librephotos.env .env\n```\n\nEdit the `.env` file and set these critical variables:\n\n```bash\n# Path to your photo library (for testing)\nscanDirectory=/path/to/your/test/photos\n\n# Path to LibrePhotos data\ndata=./librephotos/data\n\n# IMPORTANT: Path where you cloned the repositories\ncodedir=~/dev/librephotos\n```\n\n### Step 3: Start the Development Environment\n\n```bash\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml up -d\n```\n\nThis command:\n- Builds development images with hot-reload enabled\n- Mounts your local source code into the containers\n- Starts all required services (backend, frontend, database, proxy)\n\nAccess LibrePhotos at: **http://localhost:3000**\n\n### Rebuilding After Dependency Changes\n\nIf you add new dependencies to `requirements.txt` or `package.json`:\n\n```bash\n# Rebuild backend\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml build --no-cache backend\n\n# Rebuild frontend\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml build --no-cache frontend\n\n# Restart containers\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml up -d\n```\n\n---\n\n## Docker & Docker Compose\n\n### Architecture Overview\n\nLibrePhotos uses a microservices architecture with four main containers:\n\n| Container | Purpose |\n|-------------|------------------------------------------------------|\n| `backend` | Django API server, ML models, background jobs |\n| `frontend` | React web application |\n| `proxy` | Nginx reverse proxy, serves static files |\n| `db` | PostgreSQL database |\n\n### Useful Docker Commands\n\n```bash\n# View running containers\ndocker compose ps\n\n# View logs (all containers)\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml logs -f\n\n# View logs (specific container)\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml logs -f backend\n\n# Restart a container\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml restart backend\n\n# Stop all containers\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml down\n\n# Stop and remove volumes (fresh start)\ndocker compose -f docker-compose.yml -f docker-compose.dev.yml down -v\n\n# Execute command in container\ndocker exec -it backend bash\ndocker exec -it frontend sh\n\n# Run Django management commands\ndocker exec -it backend python manage.py migrate\ndocker exec -it backend python manage.py createsuperuser\n```\n\n### Development vs Production\n\n| Aspect | Development (`docker-compose.dev.yml`) | Production (`docker-compose.yml`) |\n|-------------------|-----------------------------------------------------|--------------------------------------|\n| Source code | Mounted from local filesystem | Built into image |\n| Hot reload | ✅ Enabled | ❌ Disabled |\n| Debug mode | ✅ `DEBUG=1` | ❌ `DEBUG=0` |\n| Build time | Longer (builds from source) | Fast (pulls pre-built images) |\n| Additional tools | pgAdmin available on port 3001 | Minimal |\n\n---\n\n## IDE Recommendations\n\n### VS Code (Recommended)\n\nVS Code is the recommended IDE with excellent Docker and Python support.\n\n**Recommended Extensions:**\n- **Python** - Python language support\n- **Pylance** - Fast Python language server\n- **Docker** - Docker container management\n- **Remote - Containers** - Develop inside Docker containers\n- **ESLint** - JavaScript/TypeScript linting\n- **Prettier** - Code formatting\n\n**Workspace Settings:**\n\nThe repository includes VS Code settings in `librephotos-docker/vscode/settings.json` that are automatically mounted into the backend container.\n\n**Attaching to Backend Container:**\n\nFor the best development experience, you can attach VS Code directly to the running backend container:\n\n1. Install the \"Remote - Containers\" extension\n2. Open Command Palette (`Ctrl+Shift+P`)\n3. Run \"Remote-Containers: Attach to Running Container\"\n4. Select the `backend` container\n5. Open the `/code` folder\n\n### PyCharm\n\nPyCharm Professional supports Docker interpreters natively:\n\n1. Go to Settings → Project → Python Interpreter\n2. Add Interpreter → On Docker Compose\n3. Select the `docker-compose.yml` and `docker-compose.dev.yml` files\n4. Choose the `backend` service\n\n### Other IDEs\n\nAny IDE with Python and TypeScript support will work. Key requirements:\n- Python 3.11+ interpreter support\n- ESLint/Prettier integration for frontend\n- Docker integration (optional but helpful)\n\n---\n\n## Code Quality Standards\n\n### Backend (Python/Django)\n\n**Linting and Formatting:**\n\nWe use `ruff` for linting and formatting (configured in `pyproject.toml`):\n\n```bash\n# Inside the backend container\ncd /code\npip install ruff\nruff check .\nruff format .\n```\n\n**Pre-commit Hooks:**\n\nInstall pre-commit hooks for automatic formatting:\n\n```bash\npip install pre-commit\npre-commit install\n```\n\n**Code Style:**\n- Line length: 88 characters\n- Use type hints where practical\n- Follow PEP 8 naming conventions\n- Write docstrings for public functions\n\n### Frontend (React/TypeScript)\n\n**Linting and Formatting:**\n\n```bash\n# Inside frontend container or locally\nyarn lint:error # Check for errors\nyarn lint:warning:fix # Fix linting issues\n```\n\n**Code Style:**\n- Line length: 120 characters\n- Use Prettier for formatting (configured in `prettier.config.cjs`)\n- Prefer TypeScript types over interfaces (project convention)\n- Use functional components with hooks\n- Follow the slice pattern for Redux state management\n\n### Pull Request Checklist\n\nBefore submitting a PR, ensure:\n\n- [ ] Code follows the project's style guidelines\n- [ ] All linting passes without errors\n- [ ] New features include tests (if applicable)\n- [ ] Documentation is updated (if needed)\n- [ ] Commit messages are clear and descriptive\n- [ ] The PR addresses a single concern/feature\n\n---\n\n## How to Open a Pull Request\n\n### Step 1: Fork the Repository\n\n1. Navigate to the repository you want to contribute to on GitHub\n2. Click the \"Fork\" button in the top right corner\n3. Clone your fork locally:\n\n```bash\ngit clone https://github.com/YOUR-USERNAME/librephotos.git\ncd librephotos\ngit remote add upstream https://github.com/LibrePhotos/librephotos.git\n```\n\n### Step 2: Create a Feature Branch\n\nAlways create a new branch for your work:\n\n```bash\ngit checkout -b feature/my-awesome-feature\n# or\ngit checkout -b fix/bug-description\n```\n\n### Step 3: Make Your Changes\n\n1. Write your code following the code quality standards above\n2. Test your changes thoroughly\n3. Commit your changes with descriptive messages:\n\n```bash\ngit add .\ngit commit -m \"feat: add support for XYZ\"\n# or\ngit commit -m \"fix: resolve issue with ABC\"\n```\n\n**Commit Message Guidelines:**\n- Use present tense (\"add feature\" not \"added feature\")\n- Keep the first line under 72 characters\n- Reference issues when applicable: `fix: resolve login bug (#123)`\n\n### Step 4: Push and Create Pull Request\n\n```bash\ngit push origin feature/my-awesome-feature\n```\n\nThen on GitHub:\n1. Navigate to your fork\n2. Click \"Compare & pull request\"\n3. Fill out the PR template with:\n - Clear description of changes\n - Reference to related issues\n - Screenshots (for UI changes)\n - Testing instructions\n\n### Step 5: Respond to Review\n\n- Address reviewer feedback promptly\n- Make requested changes in new commits\n- Be open to suggestions and discussion\n\n---\n\n## Getting Help\n\n- **Discord:** [Join our Discord server](https://discord.gg/xwRvtSDGWb)\n- **GitHub Issues:** [Report bugs or request features](https://github.com/LibrePhotos/librephotos/issues)\n- **Documentation:** [docs.librephotos.com](https://docs.librephotos.com)\n- **Development Videos:** [Niaz Faridani-Rad's YouTube channel](https://www.youtube.com/channel/UCZJ2pk2BPKxwbuCV9LWDR0w)\n\n### Debugging Tips\n\n**Backend (Django):**\n\nUse `pdb` for debugging:\n\n```python\nimport pdb; pdb.set_trace()\n```\n\nThen attach to the container:\n\n```bash\ndocker attach $(docker ps --filter name=backend -q)\n```\n\nPress `Ctrl+P` followed by `Ctrl+Q` to detach without stopping the container.\n\n**Frontend (React):**\n\n- Use React DevTools browser extension\n- Use Redux DevTools for state debugging\n- Enable WDYR by setting `WDYR=True` in your `.env`\n\n**API Documentation:**\n\nAfter starting LibrePhotos, access the API docs at:\n- Swagger: http://localhost:3000/api/swagger\n- ReDoc: http://localhost:3000/api/redoc\n\n---\n\n## License\n\nBy contributing to LibrePhotos, you agree that your contributions will be licensed under the MIT License.\n\nThank you for contributing! 🎉\n\n" }, { "path": "LICENSE", "content": "MIT License\n\nCopyright (c) 2017 Hooram Nam\n\nPermission is hereby granted, free of charge, to any person obtaining a copy\nof this software and associated documentation files (the \"Software\"), to deal\nin the Software without restriction, including without limitation the rights\nto use, copy, modify, merge, publish, distribute, sublicense, and/or sell\ncopies of the Software, and to permit persons to whom the Software is\nfurnished to do so, subject to the following conditions:\n\nThe above copyright notice and this permission notice shall be included in all\ncopies or substantial portions of the Software.\n\nTHE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\nIMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\nFITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE\nAUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\nLIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,\nOUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE\nSOFTWARE.\n" }, { "path": "README.md", "content": "[![Discord](https://img.shields.io/discord/784619049208250388?style=plastic)][discord] [![Website](https://img.shields.io/website?down_color=lightgrey&down_message=offline&style=plastic&up_color=blue&up_message=online&url=https%3A%2F%2Flibrephotos.com)](https://librephotos.com/)\n[![Read the docs](https://img.shields.io/static/v1?label=Read&message=the%20docs&color=blue&style=plastic)](https://docs.librephotos.com/) [![GitHub contributors](https://img.shields.io/github/contributors/librephotos/librephotos?style=plastic)](https://github.com/LibrePhotos/librephotos/graphs/contributors) [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg?style=plastic)](https://github.com/LibrePhotos/librephotos/blob/dev/LICENSE)\n\n\"Translation\n\n\n# LibrePhotos\n\n![](https://github.com/LibrePhotos/librephotos/blob/dev/screenshots/mockups_main_fhd.png?raw=true)\nMockup designed by rawpixel.com / Freepik\n\nA self-hosted, open-source photo management service with automatic face recognition, object detection, and semantic search — powered by modern machine learning.\n\n- **Stable** demo is available here: https://demo1.librephotos.com/ . User is ```demo```, password is ```demo1234``` (with sample images).\n- Latest **development** demo is available here: https://demo2.librephotos.com/ (same user/password)\n- You can watch development videos on [Niaz Faridani-Rad's channel](https://www.youtube.com/channel/UCZJ2pk2BPKxwbuCV9LWDR0w)\n- You can join our [Discord][discord].\n\n## Installation\n\nStep-by-step installation instructions are available in our [documentation](https://docs.librephotos.com/docs/installation/standard-install).\n\n### System Requirements\n\n| Resource | Minimum | Recommended |\n|----------|---------|-------------|\n| RAM | 4 GB | 8 GB+ |\n| Storage | 10 GB (plus your photo library) | SSD recommended |\n| CPU | 2 cores | 4+ cores |\n| OS | Any Docker-compatible OS | Linux |\n\n> **Note:** Machine learning features (face recognition, scene classification, image captioning) are memory-intensive. 8 GB+ RAM is strongly recommended for smooth operation.\n\n## Features\n\n - Support for all types of photos including raw photos\n - Support for videos\n - Timeline view\n - Scans pictures on the file system\n - Multiuser support\n - Generate albums based on events like \"Thursday in Berlin\"\n - Face recognition / Face classification\n - Reverse geocoding\n - Object / Scene detection\n - Semantic image search\n - Search by metadata\n\n## Tech Stack\n\n### Backend\n\n- **Framework:** [Django 5](https://www.djangoproject.com/) with [Django REST Framework](https://www.django-rest-framework.org/)\n- **Database:** [PostgreSQL](https://www.postgresql.org/)\n- **Task Queue:** [Django-Q2](https://github.com/django-q2/django-q2)\n- **Image Conversion:** [ImageMagick](https://github.com/ImageMagick/ImageMagick)\n- **Video Conversion:** [FFmpeg](https://github.com/FFmpeg/FFmpeg)\n- **Exif Support:** [ExifTool](https://github.com/exiftool/exiftool)\n\n### Frontend\n\n- **UI:** [React 18](https://react.dev/) with [TypeScript](https://www.typescriptlang.org/)\n- **Build Tool:** [Vite](https://vite.dev/)\n- **Component Library:** [Mantine](https://mantine.dev/)\n- **Routing:** [TanStack Router](https://tanstack.com/router)\n- **Data Fetching:** [TanStack Query](https://tanstack.com/query)\n- **Maps:** [MapLibre GL](https://maplibre.org/)\n- **Internationalization:** [i18next](https://www.i18next.com/)\n\n### Machine Learning\n\n- **Face detection:** [face_recognition](https://github.com/ageitgey/face_recognition)\n- **Face classification/clustering:** [scikit-learn](https://scikit-learn.org/) and [hdbscan](https://github.com/scikit-learn-contrib/hdbscan)\n- **Image captioning:** [im2txt](https://github.com/HughKu/Im2txt)\n- **Scene classification:** [places365](http://places.csail.mit.edu/)\n- **Reverse geocoding:** [geopy](https://github.com/geopy/geopy)\n\n### Infrastructure\n\n- **Deployment:** [Docker](https://www.docker.com/) & [Docker Compose](https://docs.docker.com/compose/)\n- **Reverse Proxy:** [Nginx](https://nginx.org/)\n\n### API Documentation\n\nAfter starting LibrePhotos, interactive API docs are available at:\n\n- **Swagger UI:** `http://localhost:3000/api/swagger`\n- **ReDoc:** `http://localhost:3000/api/redoc`\n\n## How to help out\n\n- ⭐ **Star** this repository if you like this project!\n- 🚀 **Developing**: Get started in less than 30 minutes by following [this guide](https://docs.librephotos.com/docs/development/dev-install). Also see our [CONTRIBUTING.md](CONTRIBUTING.md) for detailed development setup, code quality standards, and PR guidelines.\n- 🗒️ **Documentation**: Improving the documentation is as simple as submitting a pull request [here](https://github.com/LibrePhotos/librephotos.docs)\n- 🧪 **Testing**: If you want to help find bugs, use the ```dev``` tag and update it regularly. If you find a bug, open an issue.\n- 🧑‍🤝‍🧑 **Outreach**: Talk about this project with other people and help them to get started too!\n- 🌐 **Translations**: Make LibrePhotos accessible to more people with [weblate](https://hosted.weblate.org/engage/librephotos/).\n- 💸 [**Donate**](https://github.com/sponsors/derneuere) to the developers of LibrePhotos\n\n## Related Projects\n\n| Repository | Description |\n|------------|-------------|\n| [librephotos-frontend](https://github.com/LibrePhotos/librephotos-frontend) | React/TypeScript web frontend |\n| [librephotos-docker](https://github.com/LibrePhotos/librephotos-docker) | Docker Compose deployment configurations |\n| [librephotos.docs](https://github.com/LibrePhotos/librephotos.docs) | Documentation website source |\n| [librephotos-mobile](https://github.com/LibrePhotos/librephotos-mobile) | Mobile application |\n\n## License\n\nThis project is licensed under the [MIT License](LICENSE).\n\n[discord]: https://discord.gg/xwRvtSDGWb\n" }, { "path": "api/__init__.py", "content": "default_app_config = \"api.apps.ApiConfig\"\n" }, { "path": "api/admin.py", "content": "from django.contrib import admin\nfrom django_q.tasks import AsyncTask\n\nfrom .models import (\n AlbumAuto,\n AlbumDate,\n AlbumPlace,\n AlbumThing,\n AlbumUser,\n Cluster,\n Face,\n File,\n LongRunningJob,\n Person,\n Photo,\n User,\n Thumbnail,\n)\n\n\ndef deduplicate_faces_function(queryset):\n for photo in queryset:\n # Get all faces in the photo\n faces = Face.objects.filter(photo=photo)\n # Check if there are any faces which have similar bounding boxes\n for face in faces:\n margin = int((face.location_right - face.location_left) * 0.05)\n similar_faces = Face.objects.filter(\n photo=photo,\n location_top__lte=face.location_top + margin,\n location_top__gte=face.location_top - margin,\n location_right__lte=face.location_right + margin,\n location_right__gte=face.location_right - margin,\n location_bottom__lte=face.location_bottom + margin,\n location_bottom__gte=face.location_bottom - margin,\n location_left__lte=face.location_left + margin,\n location_left__gte=face.location_left - margin,\n )\n if len(similar_faces) > 1:\n # Divide between faces with a person label and faces without\n faces_with_person_label = []\n faces_without_person_label = []\n for similar_face in similar_faces:\n if similar_face.person:\n faces_with_person_label.append(similar_face)\n else:\n faces_without_person_label.append(similar_face)\n # If there are faces with a person label, keep the first one and delete the rest\n for similar_face in faces_with_person_label[1:]:\n similar_face.delete()\n # If there are faces with a person label, delete all of them\n if len(faces_with_person_label) > 0:\n for similar_face in faces_without_person_label:\n similar_face.delete()\n # Otherwise, keep the first face and delete the rest\n else:\n for similar_face in faces_without_person_label[1:]:\n similar_face.delete()\n\n\n@admin.register(Face)\nclass FaceAdmin(admin.ModelAdmin):\n list_display = (\n \"id\",\n \"cluster_person\",\n \"cluster_probability\",\n \"classification_person\",\n \"cluster\",\n \"photo\",\n \"person\",\n )\n list_filter = (\"person\", \"cluster\")\n\n\n@admin.register(Photo)\nclass PhotoAdmin(admin.ModelAdmin):\n actions = [\"deduplicate_faces\"]\n list_display = [\n \"image_hash\",\n \"owner\",\n \"main_file\",\n \"last_modified\",\n \"added_on\",\n \"size\",\n ]\n list_filter = [\"owner\"]\n\n def deduplicate_faces(self, request, queryset):\n AsyncTask(\n deduplicate_faces_function,\n queryset=queryset,\n ).run()\n\n\n@admin.register(Thumbnail)\nclass ThumbnailAdmin(admin.ModelAdmin):\n list_display = [\"photo\", \"aspect_ratio\"]\n raw_id_fields = [\"photo\"]\n\n\nadmin.site.register(Person)\nadmin.site.register(AlbumAuto)\nadmin.site.register(AlbumUser)\nadmin.site.register(AlbumThing)\nadmin.site.register(AlbumDate)\nadmin.site.register(AlbumPlace)\nadmin.site.register(Cluster)\nadmin.site.register(LongRunningJob)\nadmin.site.register(File)\nadmin.site.register(User)\n" }, { "path": "api/all_tasks.py", "content": "import io\nimport os\nimport zipfile\n\nfrom django.conf import settings\nfrom django.utils import timezone\nfrom django_q.tasks import AsyncTask, schedule\n\nfrom api import util\nfrom api.models.long_running_job import LongRunningJob\n\n\ndef create_download_job(job_type, user, photos, filename):\n lrj = LongRunningJob.create_job(\n user=user,\n job_type=job_type,\n )\n if job_type == LongRunningJob.JOB_DOWNLOAD_PHOTOS:\n AsyncTask(\n zip_photos_task, job_id=lrj.job_id, user=user, photos=photos, filename=filename\n ).run()\n\n return lrj.job_id\n\n\ndef zip_photos_task(job_id, user, photos, filename):\n lrj = LongRunningJob.objects.get(job_id=job_id)\n lrj.start()\n count = len(photos)\n lrj.update_progress(current=0, target=count)\n output_directory = os.path.join(settings.MEDIA_ROOT, \"zip\")\n zip_file_name = filename\n done_count = 0\n try:\n if not os.path.exists(output_directory):\n os.mkdir(output_directory)\n mf = io.BytesIO()\n files_added = {} # Track files by path to avoid duplicates\n\n for photo in photos:\n done_count = done_count + 1\n\n # Collect all files for this photo.\n # NOTE: main_file is not guaranteed to be included in Photo.files.\n all_files = []\n if getattr(photo, \"main_file\", None) is not None:\n all_files.append(photo.main_file)\n all_files.extend(list(photo.files.all()))\n\n # Back-compat: some datasets may still represent RAW+JPEG / Live Photo variants\n # as deprecated stacks. Include those stack members' files too.\n try:\n variant_stacks = photo.stacks.filter(\n stack_type__in=[\"raw_jpeg\", \"live_photo\"]\n ).prefetch_related(\"photos\", \"photos__files\", \"photos__main_file\")\n for stack in variant_stacks:\n for stack_photo in stack.photos.all():\n if getattr(stack_photo, \"main_file\", None) is not None:\n all_files.append(stack_photo.main_file)\n all_files.extend(list(stack_photo.files.all()))\n except Exception:\n # If stacks aren't available for some reason, just proceed with variants.\n pass\n\n # Include embedded media variants for every collected file (not just main_file)\n for file_obj in list(all_files):\n try:\n if file_obj and file_obj.embedded_media.exists():\n all_files.extend(list(file_obj.embedded_media.all()))\n except Exception:\n continue\n \n # Add each file to the zip\n for file_obj in all_files:\n if not file_obj or not file_obj.path:\n continue\n \n # Skip if file doesn't exist on disk\n if not os.path.exists(file_obj.path):\n util.logger.warning(f\"File not found, skipping: {file_obj.path}\")\n continue\n \n # Skip if already added (avoid duplicates)\n if file_obj.path in files_added:\n continue\n \n file_name = os.path.basename(file_obj.path)\n \n # Handle duplicate filenames in the zip\n if file_name in files_added.values():\n # Find a unique name by prepending a counter\n counter = 1\n base_name, ext = os.path.splitext(file_name)\n while f\"{base_name}_{counter}{ext}\" in files_added.values():\n counter += 1\n file_name = f\"{base_name}_{counter}{ext}\"\n \n files_added[file_obj.path] = file_name\n \n with zipfile.ZipFile(mf, mode=\"a\", compression=zipfile.ZIP_DEFLATED) as zf:\n zf.write(file_obj.path, arcname=file_name)\n \n lrj.update_progress(current=done_count, target=count)\n \n with open(os.path.join(output_directory, zip_file_name), \"wb\") as output_file:\n output_file.write(mf.getvalue())\n\n except Exception as e:\n util.logger.error(f\"Error while converting files to zip: {e}\")\n\n lrj.complete()\n # scheduling a task to delete the zip file after a day\n execution_time = timezone.now() + timezone.timedelta(days=1)\n schedule(\"api.all_tasks.delete_zip_file\", filename, next_run=execution_time)\n return os.path.join(output_directory, zip_file_name)\n\n\ndef delete_zip_file(filename):\n file_path = os.path.join(settings.MEDIA_ROOT, \"zip\", filename)\n try:\n if not os.path.exists(file_path):\n util.logger.error(f\"Error while deleting file not found at : {file_path}\")\n return\n else:\n os.remove(file_path)\n util.logger.info(f\"file deleted sucessfully at path : {file_path}\")\n return\n\n except Exception as e:\n util.logger.error(f\"Error while deleting file: {e}\")\n return e\n" }, { "path": "api/api_util.py", "content": "import os\nimport random\nimport stat\n\nfrom api.models import (\n LongRunningJob,\n Photo,\n)\nfrom api.serializers.job import LongRunningJobSerializer\nfrom api.util import logger\n\n\ndef get_current_job():\n job_detail = None\n running_job = (\n LongRunningJob.objects.filter(finished=False).order_by(\"-started_at\").first()\n )\n if running_job:\n job_detail = LongRunningJobSerializer(running_job).data\n return job_detail\n\n\ndef shuffle(list):\n random.shuffle(list)\n return list\n\n\ndef is_hidden(filepath):\n name = os.path.basename(os.path.abspath(filepath))\n return name.startswith(\".\") or has_hidden_attribute(filepath)\n\n\ndef has_hidden_attribute(filepath):\n try:\n return bool(os.stat(filepath).st_file_attributes & stat.FILE_ATTRIBUTE_HIDDEN)\n except Exception:\n return False\n\n\ndef path_to_dict(path, recurse=2):\n d = {\"title\": os.path.basename(path), \"absolute_path\": path}\n if recurse > 0:\n d[\"children\"] = [\n path_to_dict(os.path.join(path, x), recurse - 1)\n for x in os.scandir(path)\n if os.path.isdir(os.path.join(path, x))\n and not is_hidden(os.path.join(path, x))\n ]\n else:\n d[\"children\"] = []\n # sort children by title alphabetically (case insensitive)\n d[\"children\"] = sorted(d[\"children\"], key=lambda k: k[\"title\"].lower())\n return d\n\n\ndef get_search_term_examples(user):\n default_search_terms = [\n \"for people\",\n \"for places\",\n \"for things\",\n \"for time\",\n \"for file path or file name\",\n ]\n\n possible_ids = list(\n Photo.objects.filter(owner=user)\n .exclude(caption_instance__captions_json={})\n .exclude(caption_instance__captions_json__isnull=True)[:1000]\n .values_list(\"image_hash\", flat=True)\n )\n if len(possible_ids) > 99:\n possible_ids = random.choices(possible_ids, k=100)\n logger.info(f\"{len(possible_ids)} possible ids\")\n try:\n samples = (\n Photo.objects.filter(owner=user)\n .exclude(caption_instance__captions_json={})\n .exclude(caption_instance__captions_json__isnull=True)\n .filter(image_hash__in=possible_ids)\n .prefetch_related(\"faces\")\n .prefetch_related(\"faces__person\")\n .prefetch_related(\"caption_instance\")\n .all()\n )\n except ValueError:\n return default_search_terms\n\n search_data = []\n search_terms = default_search_terms\n logger.info(\"Getting search terms for user %s\", user.id)\n logger.info(\"Found %s photos\", len(samples))\n for p in samples:\n faces = p.faces.all()\n terms_loc = \"\"\n if (\n p.geolocation_json\n and p.geolocation_json != {}\n and \"features\" in p.geolocation_json\n ):\n terms_loc = [\n f[\"text\"]\n for f in p.geolocation_json[\"features\"][-5:]\n if \"text\" in f and not f[\"text\"].isdigit()\n ]\n terms_time = \"\"\n if p.exif_timestamp:\n terms_time = [str(p.exif_timestamp.year)]\n terms_people = []\n if p.faces.count() > 0:\n terms_people = [\n f.person.name.split(\" \")[0] if f.person else \"\" for f in faces\n ]\n terms_things = \"\"\n if (\n p.caption_instance\n and p.caption_instance.captions_json\n and p.caption_instance.captions_json.get(\"places365\") is not None\n ):\n terms_things = p.caption_instance.captions_json[\"places365\"][\"categories\"]\n\n terms = {\n \"loc\": terms_loc,\n \"time\": terms_time,\n \"people\": terms_people,\n \"things\": terms_things,\n }\n\n search_data.append(terms)\n search_terms = []\n for datum in search_data:\n term_time = \"\"\n term_thing = \"\"\n term_loc = \"\"\n term_people = \"\"\n if datum[\"loc\"]:\n term_loc = random.choice(datum[\"loc\"])\n search_terms.append(term_loc)\n if datum[\"time\"]:\n term_time = random.choice(datum[\"time\"])\n search_terms.append(term_time)\n if datum[\"things\"]:\n term_thing = random.choice(datum[\"things\"])\n search_terms.append(term_thing)\n if datum[\"people\"]:\n term_people = random.choice(datum[\"people\"])\n search_terms.append(term_people)\n\n search_term_loc_people = \" \".join(shuffle([term_loc, term_people]))\n if random.random() > 0.3:\n search_terms.append(search_term_loc_people)\n\n search_term_time_people = \" \".join(shuffle([term_time, term_people]))\n if random.random() > 0.3:\n search_terms.append(search_term_time_people)\n\n search_term_people_thing = \" \".join(shuffle([term_people, term_thing]))\n if random.random() > 0.9:\n search_terms.append(search_term_people_thing)\n\n search_term_all = \" \".join(\n shuffle([term_loc, term_people, term_time, term_thing])\n )\n if random.random() > 0.95:\n search_terms.append(search_term_all)\n\n search_term_loc_time = \" \".join(shuffle([term_loc, term_time]))\n if random.random() > 0.3:\n search_terms.append(search_term_loc_time)\n\n search_term_loc_thing = \" \".join(shuffle([term_loc, term_thing]))\n if random.random() > 0.9:\n search_terms.append(search_term_loc_thing)\n\n search_term_time_thing = \" \".join(shuffle([term_time, term_thing]))\n if random.random() > 0.9:\n search_terms.append(search_term_time_thing)\n\n return list(filter(lambda x: len(x), set([x.strip() for x in search_terms])))\n" }, { "path": "api/apps.py", "content": "from django.apps import AppConfig\n\n\nclass ApiConfig(AppConfig):\n name = \"api\"\n verbose_name = \"LibrePhotos\"\n" }, { "path": "api/autoalbum.py", "content": "from datetime import datetime, timedelta\n\nimport numpy as np\nimport pytz\nfrom django.db.models import Q\n\nfrom api.models import (\n AlbumAuto,\n AlbumDate,\n AlbumPlace,\n AlbumThing,\n AlbumUser,\n Face,\n File,\n LongRunningJob,\n Photo,\n)\nfrom api.util import logger\n\n\ndef regenerate_event_titles(user, job_id):\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_GENERATE_AUTO_ALBUM_TITLES,\n job_id=job_id,\n )\n try:\n aus = AlbumAuto.objects.filter(owner=user).prefetch_related(\"photos\")\n target_count = len(aus)\n for idx, au in enumerate(aus):\n logger.info(f\"job {job_id}: {idx}\")\n au._generate_title()\n au.save()\n lrj.update_progress(current=idx + 1, target=target_count)\n\n lrj.complete()\n logger.info(f\"job {job_id}: updated lrj entry to db\")\n\n except Exception as e:\n logger.exception(\"An error occurred\")\n lrj.fail(error=e)\n\n return 1\n\n\ndef generate_event_albums(user, job_id):\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_GENERATE_AUTO_ALBUMS,\n job_id=job_id,\n )\n\n try:\n photos = (\n Photo.objects.filter(Q(owner=user))\n .exclude(Q(exif_timestamp=None))\n .only(\"exif_timestamp\")\n )\n\n def group(photos, dt=timedelta(hours=6)):\n photos_with_timestamp = sorted(photos, key=lambda p: p.exif_timestamp)\n groups = []\n for idx, photo in enumerate(photos_with_timestamp):\n if len(groups) == 0:\n groups.append([])\n groups[-1].append(photo)\n # Photos are sorted by timestamp, so we can just check the last photo of the last group\n # to see if it is within the time delta\n elif photo.exif_timestamp - groups[-1][-1].exif_timestamp < dt:\n groups[-1].append(photo)\n # If the photo is not within the time delta, we create a new group\n else:\n groups.append([])\n groups[-1].append(photo)\n return groups\n\n # Group images that are on the same 1 day and 12 hours interval\n groups = group(photos, dt=timedelta(days=1, hours=12))\n target_count = len(groups)\n logger.info(\n f\"job {job_id}: made {target_count} groups out of {len(photos)} images\"\n )\n\n album_locations = []\n\n date_format = \"%Y:%m:%d %H:%M:%S\"\n for idx, group in enumerate(groups):\n key = group[0].exif_timestamp - timedelta(hours=11, minutes=59)\n lastKey = group[-1].exif_timestamp + timedelta(hours=11, minutes=59)\n logger.info(str(key.date) + \" - \" + str(lastKey.date))\n logger.info(\n f\"job {job_id}: processing auto album with date: \"\n + key.strftime(date_format)\n + \" to \"\n + lastKey.strftime(date_format)\n )\n items = group\n if len(group) >= 2:\n qs = AlbumAuto.objects.filter(owner=user).filter(\n timestamp__range=(key, lastKey)\n )\n if qs.count() == 0:\n album = AlbumAuto(\n created_on=datetime.utcnow().replace(tzinfo=pytz.utc),\n owner=user,\n )\n album.timestamp = key\n album.save()\n\n logger.info(f\"job {job_id}: generate auto album {album.id}\")\n locs = []\n for item in items:\n album.photos.add(item)\n item.save()\n if item.exif_gps_lat and item.exif_gps_lon:\n locs.append([item.exif_gps_lat, item.exif_gps_lon])\n if len(locs) > 0:\n album_location = np.mean(np.array(locs), 0)\n album_locations.append(album_location)\n album.gps_lat = album_location[0]\n album.gps_lon = album_location[1]\n else:\n album_locations.append([])\n album._generate_title()\n album.save()\n continue\n if qs.count() == 1:\n album = qs.first()\n logger.info(f\"job {job_id}: update auto album {album.id}\")\n for item in items:\n if item in album.photos.all():\n continue\n album.photos.add(item)\n item.save()\n album._generate_title()\n album.save()\n continue\n if qs.count() > 1:\n # To-Do: Merge both auto albums\n logger.info(\n f\"job {job_id}: found multiple auto albums for date {key.strftime(date_format)}\"\n )\n continue\n\n lrj.update_progress(current=idx + 1, target=target_count)\n\n lrj.complete()\n\n except Exception as e:\n logger.exception(\"An error occurred\")\n lrj.fail(error=e)\n\n return 1\n\n\n# To-Do: This does not belong here\ndef delete_missing_photos(user, job_id):\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_DELETE_MISSING_PHOTOS,\n job_id=job_id,\n )\n try:\n missing_photos = Photo.objects.filter(\n Q(owner=user) & Q(files=None) | Q(main_file=None)\n )\n for missing_photo in missing_photos:\n album_dates = AlbumDate.objects.filter(photos=missing_photo)\n for album_date in album_dates:\n album_date.photos.remove(missing_photo)\n album_things = AlbumThing.objects.filter(photos=missing_photo)\n for album_thing in album_things:\n album_thing.photos.remove(missing_photo)\n album_places = AlbumPlace.objects.filter(photos=missing_photo)\n for album_place in album_places:\n album_place.photos.remove(missing_photo)\n album_users = AlbumUser.objects.filter(photos=missing_photo)\n for album_user in album_users:\n album_user.photos.remove(missing_photo)\n faces = Face.objects.filter(photo=missing_photo)\n faces.delete()\n # To-Do: Remove thumbnails\n\n missing_photos.delete()\n\n missing_files = File.objects.filter(Q(hash__endswith=user) & Q(missing=True))\n missing_files.delete()\n\n lrj.complete()\n except Exception as e:\n logger.exception(\"An error occurred\")\n lrj.fail(error=e)\n return 1\n" }, { "path": "api/background_tasks.py", "content": "from tqdm import tqdm\nfrom django.db import models\n\nfrom api.models import Photo\nfrom api.models.photo_caption import PhotoCaption\nfrom api.util import logger\n\n\ndef generate_captions(overwrite=False):\n if overwrite:\n photos = Photo.objects.all()\n else:\n # Find photos that don't have search captions in PhotoSearch model\n photos = Photo.objects.filter(\n models.Q(search_instance__isnull=True)\n | models.Q(search_instance__search_captions__isnull=True)\n )\n logger.info(\"%d photos to be processed for caption generation\" % photos.count())\n for photo in photos:\n logger.info(\"generating captions for %s\" % photo.main_file.path)\n caption_instance, created = PhotoCaption.objects.get_or_create(photo=photo)\n caption_instance.generate_tag_captions()\n photo.save()\n\n\ndef geolocate(overwrite=False):\n if overwrite:\n photos = Photo.objects.all()\n else:\n photos = Photo.objects.filter(geolocation_json={})\n logger.info(\"%d photos to be geolocated\" % photos.count())\n for photo in photos:\n try:\n logger.info(\"geolocating %s\" % photo.main_file.path)\n photo._geolocate()\n photo._add_location_to_album_dates()\n except Exception:\n logger.exception(\"could not geolocate photo: %s\", photo)\n\n\ndef add_photos_to_album_things():\n photos = Photo.objects.all()\n for photo in tqdm(photos):\n photo._add_to_album_place()\n" }, { "path": "api/batch_jobs.py", "content": "import os\n\nfrom django.db.models import Q\n\nfrom api import util\nfrom api.image_similarity import build_image_similarity_index\nfrom api.models.long_running_job import LongRunningJob\nfrom api.models.photo import Photo\nfrom api.semantic_search import create_clip_embeddings\n\n\ndef batch_calculate_clip_embedding(user):\n import torch\n\n lrj = LongRunningJob.create_job(\n user=user,\n job_type=LongRunningJob.JOB_CALCULATE_CLIP_EMBEDDINGS,\n start_now=True,\n )\n\n count = Photo.objects.filter(\n Q(owner=user) & Q(clip_embeddings__isnull=True)\n ).count()\n lrj.update_progress(current=0, target=count)\n \n if not torch.cuda.is_available():\n num_threads = 1\n torch.set_num_threads(num_threads)\n os.environ[\"OMP_NUM_THREADS\"] = str(num_threads)\n else:\n torch.multiprocessing.set_start_method(\"spawn\", force=True)\n\n BATCH_SIZE = 64\n util.logger.info(f\"Using threads: {torch.get_num_threads()}\")\n\n done_count = 0\n while done_count < count:\n try:\n objs = list(\n Photo.objects.filter(Q(owner=user) & Q(clip_embeddings__isnull=True))[\n :BATCH_SIZE\n ]\n )\n done_count += len(objs)\n\n if len(objs) == 0:\n break\n valid_objs = []\n for obj in objs:\n # Thumbnail could have been deleted\n if obj.thumbnail.thumbnail_big and os.path.exists(\n obj.thumbnail.thumbnail_big.path\n ):\n valid_objs.append(obj)\n imgs = list(map(lambda obj: obj.thumbnail.thumbnail_big.path, valid_objs))\n if len(valid_objs) == 0:\n continue\n\n imgs_emb, magnitudes = create_clip_embeddings(imgs)\n\n for obj, img_emb, magnitude in zip(valid_objs, imgs_emb, magnitudes):\n obj.clip_embeddings = img_emb.tolist()\n obj.clip_embeddings_magnitude = magnitude\n obj.save()\n except Exception as e:\n util.logger.error(f\"Error calculating clip embeddings: {e}\")\n\n lrj.update_progress(current=done_count, target=count)\n\n build_image_similarity_index(user)\n lrj.complete()\n" }, { "path": "api/burst_detection_rules.py", "content": "\"\"\"\nBurst detection rules module for grouping photos taken in rapid succession.\n\nThis module provides a rules-based system for detecting burst sequences,\nfollowing the same pattern as date_time_extractor.py. Rules are stored\nas JSON in each user's profile and applied sequentially.\n\nRules are divided into two categories:\n- Hard criteria: Deterministic detection based on EXIF data and filename patterns\n (e.g., BurstMode tag, SequenceNumber, filename patterns like IMG_001_BURST001)\n- Soft criteria: Estimation based on timestamp proximity and visual similarity\n (e.g., photos within 2 seconds of each other from the same camera)\n\nBy default, only hard criteria rules are enabled.\n\"\"\"\n\nimport json\nimport os\nimport re\nfrom datetime import timedelta\n\nfrom api.metadata.tags import Tags\nfrom api.util import logger\n\n\nclass BurstRuleTypes:\n \"\"\"Types of burst detection rules.\"\"\"\n\n # Hard criteria (deterministic)\n EXIF_BURST_MODE = \"exif_burst_mode\"\n EXIF_SEQUENCE_NUMBER = \"exif_sequence_number\"\n FILENAME_PATTERN = \"filename_pattern\"\n\n # Soft criteria (estimation)\n TIMESTAMP_PROXIMITY = \"timestamp_proximity\"\n VISUAL_SIMILARITY = \"visual_similarity\"\n\n\nclass BurstRuleCategory:\n \"\"\"Categories for burst detection rules.\"\"\"\n\n HARD = \"hard\" # Deterministic (EXIF, filenames)\n SOFT = \"soft\" # Estimation (timestamps, visual similarity)\n\n\n# Predefined filename patterns for burst detection\nBURST_FILENAME_PATTERNS = {\n # Pattern name: (regex, description)\n \"burst_suffix\": (\n r\"_BURST\\d+\",\n \"Files with _BURST followed by numbers (e.g., IMG_001_BURST001.jpg)\",\n ),\n \"sequence_suffix\": (\n r\"_\\d{3,}$\",\n \"Files ending with 3+ digit sequence number (e.g., IMG_001.jpg, IMG_002.jpg)\",\n ),\n \"bracketed_sequence\": (\n r\"\\(\\d+\\)$\",\n \"Files with bracketed numbers at end (e.g., photo (1).jpg, photo (2).jpg)\",\n ),\n \"samsung_burst\": (r\"_\\d{3}_COVER\", \"Samsung burst cover images\"),\n \"iphone_burst\": (r\"IMG_\\d{4}_\\d+\", \"iPhone burst sequence pattern\"),\n}\n\n\nclass BurstDetectionRule:\n \"\"\"\n A rule for detecting burst sequences.\n\n Each rule has:\n - id: Unique identifier\n - name: Human-readable name\n - rule_type: One of BurstRuleTypes\n - category: 'hard' or 'soft' (for UI grouping)\n - enabled: Whether the rule is active\n - is_default: Whether this is a default rule\n - Type-specific parameters (e.g., interval_ms for timestamp_proximity)\n\n Additionally, each rule can have conditions:\n - condition_path: Regex to match full path\n - condition_filename: Regex to match filename\n - condition_exif: Format: \"TAG_NAME//regex_pattern\"\n \"\"\"\n\n def __init__(self, params):\n self.id = params.get(\"id\")\n self.name = params.get(\"name\", \"Unnamed rule\")\n self.rule_type = params[\"rule_type\"]\n self.category = params.get(\"category\", BurstRuleCategory.HARD)\n self.enabled = params.get(\"enabled\", True)\n self.is_default = params.get(\"is_default\", True)\n self.params = params\n\n def get_required_exif_tags(self):\n \"\"\"Return set of EXIF tags needed by this rule.\"\"\"\n tags = set()\n\n # Add condition tag if present\n condition_exif = self.params.get(\"condition_exif\")\n if condition_exif:\n tag_name = condition_exif.split(\"//\", maxsplit=1)[0]\n tags.add(tag_name)\n\n # Add rule-specific tags\n if self.rule_type == BurstRuleTypes.EXIF_BURST_MODE:\n tags.add(Tags.BURST_MODE)\n tags.add(Tags.CONTINUOUS_DRIVE)\n elif self.rule_type == BurstRuleTypes.EXIF_SEQUENCE_NUMBER:\n tags.add(Tags.SEQUENCE_NUMBER)\n tags.add(Tags.IMAGE_NUMBER)\n tags.add(Tags.SUBSEC_TIME_ORIGINAL)\n\n return tags\n\n def _check_condition_path(self, path):\n \"\"\"Check if path matches condition_path regex.\"\"\"\n condition = self.params.get(\"condition_path\")\n if condition:\n return re.search(condition, path) is not None\n return True\n\n def _check_condition_filename(self, path):\n \"\"\"Check if filename matches condition_filename regex.\"\"\"\n condition = self.params.get(\"condition_filename\")\n if condition:\n filename = os.path.basename(path)\n return re.search(condition, filename) is not None\n return True\n\n def _check_condition_exif(self, exif_tags):\n \"\"\"Check if EXIF tag value matches condition_exif pattern.\"\"\"\n condition = self.params.get(\"condition_exif\")\n if not condition:\n return True\n\n parts = condition.split(\"//\", maxsplit=1)\n if len(parts) != 2:\n logger.warning(f\"Invalid condition_exif format: {condition}\")\n return False\n\n tag_name, pattern = parts\n tag_value = exif_tags.get(tag_name)\n if not tag_value:\n return False\n return re.search(pattern, str(tag_value)) is not None\n\n def check_conditions(self, path, exif_tags):\n \"\"\"Check all conditions for this rule.\"\"\"\n return (\n self._check_condition_path(path)\n and self._check_condition_filename(path)\n and self._check_condition_exif(exif_tags)\n )\n\n def is_burst_photo(self, photo, exif_tags):\n \"\"\"\n Check if a photo is part of a burst sequence according to this rule.\n\n Args:\n photo: Photo model instance\n exif_tags: Dict of EXIF tag name -> value\n\n Returns:\n Tuple of (is_burst: bool, group_key: str or None)\n group_key can be used to group photos into the same burst\n \"\"\"\n if not self.enabled:\n return False, None\n\n path = photo.main_file.path if photo.main_file else \"\"\n\n if not self.check_conditions(path, exif_tags):\n return False, None\n\n if self.rule_type == BurstRuleTypes.EXIF_BURST_MODE:\n return self._check_exif_burst_mode(photo, exif_tags)\n elif self.rule_type == BurstRuleTypes.EXIF_SEQUENCE_NUMBER:\n return self._check_exif_sequence_number(photo, exif_tags)\n elif self.rule_type == BurstRuleTypes.FILENAME_PATTERN:\n return self._check_filename_pattern(photo, exif_tags)\n else:\n # Soft rules don't use is_burst_photo - they use group_by_proximity\n return False, None\n\n def _check_exif_burst_mode(self, photo, exif_tags):\n \"\"\"Check if EXIF BurstMode or ContinuousDrive indicates burst.\"\"\"\n burst_mode = exif_tags.get(Tags.BURST_MODE)\n continuous_drive = exif_tags.get(Tags.CONTINUOUS_DRIVE)\n\n # BurstMode: 1 = On (Canon, etc.)\n if burst_mode and str(burst_mode) in (\"1\", \"On\", \"True\", \"Yes\"):\n # Group by timestamp (rounded to second) + camera model\n camera = exif_tags.get(Tags.CAMERA, \"unknown\")\n timestamp = photo.exif_timestamp\n if timestamp:\n group_key = f\"burst_{camera}_{timestamp.strftime('%Y%m%d_%H%M%S')}\"\n return True, group_key\n return True, None\n\n # ContinuousDrive: Continuous, etc.\n if continuous_drive and continuous_drive.lower() in (\"continuous\", \"on\", \"1\"):\n camera = exif_tags.get(Tags.CAMERA, \"unknown\")\n timestamp = photo.exif_timestamp\n if timestamp:\n group_key = f\"burst_{camera}_{timestamp.strftime('%Y%m%d_%H%M%S')}\"\n return True, group_key\n return True, None\n\n return False, None\n\n def _check_exif_sequence_number(self, photo, exif_tags):\n \"\"\"Check if photo has sequence number indicating burst.\"\"\"\n sequence_num = exif_tags.get(Tags.SEQUENCE_NUMBER)\n\n # If we have a sequence number, it's likely part of a burst\n if sequence_num is not None:\n try:\n int(sequence_num) # Validate it's a valid number\n # Sequence numbers suggest burst\n # Group by directory + base timestamp\n camera = exif_tags.get(Tags.CAMERA, \"unknown\")\n timestamp = photo.exif_timestamp\n if timestamp:\n # Round to same second for grouping\n group_key = f\"seq_{camera}_{timestamp.strftime('%Y%m%d_%H%M%S')}\"\n return True, group_key\n return True, None\n except (ValueError, TypeError):\n pass\n\n return False, None\n\n def _check_filename_pattern(self, photo, exif_tags):\n \"\"\"Check if filename matches burst pattern.\"\"\"\n if not photo.main_file:\n return False, None\n\n filename = os.path.basename(photo.main_file.path)\n basename = os.path.splitext(filename)[0]\n\n # Get pattern type from params, default to checking all patterns\n pattern_type = self.params.get(\"pattern_type\", \"all\")\n custom_pattern = self.params.get(\"custom_pattern\")\n\n if custom_pattern:\n if re.search(custom_pattern, basename):\n # Extract base name for grouping (remove trailing numbers/burst suffix)\n base = re.sub(r\"(_BURST\\d+|_\\d{3,}|\\(\\d+\\))$\", \"\", basename)\n directory = os.path.dirname(photo.main_file.path)\n group_key = f\"filename_{directory}_{base}\"\n return True, group_key\n elif pattern_type == \"all\":\n # Check all predefined patterns\n for pattern_name, (pattern, _) in BURST_FILENAME_PATTERNS.items():\n if re.search(pattern, basename, re.IGNORECASE):\n base = re.sub(\n r\"(_BURST\\d+|_\\d{3,}|\\(\\d+\\)|_COVER)$\",\n \"\",\n basename,\n flags=re.IGNORECASE,\n )\n directory = os.path.dirname(photo.main_file.path)\n group_key = f\"filename_{directory}_{base}\"\n return True, group_key\n else:\n # Check specific pattern\n if pattern_type in BURST_FILENAME_PATTERNS:\n pattern, _ = BURST_FILENAME_PATTERNS[pattern_type]\n if re.search(pattern, basename, re.IGNORECASE):\n base = re.sub(\n r\"(_BURST\\d+|_\\d{3,}|\\(\\d+\\)|_COVER)$\",\n \"\",\n basename,\n flags=re.IGNORECASE,\n )\n directory = os.path.dirname(photo.main_file.path)\n group_key = f\"filename_{directory}_{base}\"\n return True, group_key\n\n return False, None\n\n\ndef check_filename_pattern(photo, pattern_type=\"any\"):\n \"\"\"\n Check if a photo's filename matches a burst pattern.\n\n Standalone function for testing and external use.\n\n Args:\n photo: Photo model instance with main_file\n pattern_type: \"any\" to check all patterns, or specific pattern name\n (e.g., \"burst_suffix\", \"sequence_suffix\", \"bracketed_sequence\",\n \"samsung_burst\", \"iphone_burst\")\n\n Returns:\n Tuple of (matches: bool, group_key: str or None)\n group_key can be used to group photos into the same burst\n \"\"\"\n if not photo.main_file:\n return False, None\n\n filename = os.path.basename(photo.main_file.path)\n basename = os.path.splitext(filename)[0]\n directory = os.path.dirname(photo.main_file.path)\n\n if pattern_type == \"any\" or pattern_type == \"all\":\n # Check all predefined patterns\n for pattern_name, (pattern, _) in BURST_FILENAME_PATTERNS.items():\n if re.search(pattern, basename, re.IGNORECASE):\n base = re.sub(\n r\"(_BURST\\d+|_\\d{3,}|\\(\\d+\\)|_COVER)$\",\n \"\",\n basename,\n flags=re.IGNORECASE,\n )\n group_key = f\"filename_{directory}_{base}\"\n return True, group_key\n else:\n # Check specific pattern\n if pattern_type in BURST_FILENAME_PATTERNS:\n pattern, _ = BURST_FILENAME_PATTERNS[pattern_type]\n if re.search(pattern, basename, re.IGNORECASE):\n base = re.sub(\n r\"(_BURST\\d+|_\\d{3,}|\\(\\d+\\)|_COVER)$\",\n \"\",\n basename,\n flags=re.IGNORECASE,\n )\n group_key = f\"filename_{directory}_{base}\"\n return True, group_key\n\n return False, None\n\n\ndef group_photos_by_timestamp(photos, interval_ms=2000, require_same_camera=True):\n \"\"\"\n Group photos by timestamp proximity (soft criterion).\n\n Args:\n photos: QuerySet of Photo objects ordered by exif_timestamp\n interval_ms: Maximum milliseconds between consecutive burst shots\n require_same_camera: If True, only group photos from same camera\n\n Returns:\n List of lists, each inner list is a group of Photo objects\n \"\"\"\n if not photos:\n return []\n\n interval = timedelta(milliseconds=interval_ms)\n groups = []\n current_group = []\n prev_photo = None\n prev_camera = None\n\n for photo in photos:\n if not photo.exif_timestamp:\n continue\n\n # Get camera info for same-camera check\n camera = None\n if require_same_camera and hasattr(photo, \"metadata\") and photo.metadata:\n camera = f\"{photo.metadata.camera_make or ''}_{photo.metadata.camera_model or ''}\"\n\n if prev_photo is None:\n current_group = [photo]\n prev_photo = photo\n prev_camera = camera\n continue\n\n # Check time difference\n time_diff = photo.exif_timestamp - prev_photo.exif_timestamp\n\n # Check if same camera (if required)\n same_camera = True\n if require_same_camera and camera and prev_camera:\n same_camera = camera == prev_camera\n\n if time_diff <= interval and same_camera:\n # Part of same burst\n current_group.append(photo)\n else:\n # End of current burst, save if we have a group\n if len(current_group) >= 2:\n groups.append(current_group)\n # Start new group\n current_group = [photo]\n\n prev_photo = photo\n prev_camera = camera\n\n # Don't forget the last group\n if len(current_group) >= 2:\n groups.append(current_group)\n\n return groups\n\n\ndef group_photos_by_visual_similarity(photos, similarity_threshold=15):\n \"\"\"\n Group photos by visual similarity (soft criterion).\n\n Uses perceptual hash comparison to group visually similar photos.\n\n Args:\n photos: List of Photo objects\n similarity_threshold: Maximum hamming distance (lower = more similar)\n\n Returns:\n List of lists, each inner list is a group of visually similar Photo objects\n \"\"\"\n from api.perceptual_hash import hamming_distance\n\n if not photos:\n return []\n\n # Filter photos with perceptual hashes\n photos_with_hash = [p for p in photos if p.perceptual_hash]\n\n if len(photos_with_hash) < 2:\n return []\n\n # Simple clustering: group consecutive similar photos\n groups = []\n current_group = [photos_with_hash[0]]\n\n for i in range(1, len(photos_with_hash)):\n photo = photos_with_hash[i]\n prev_photo = photos_with_hash[i - 1]\n\n distance = hamming_distance(photo.perceptual_hash, prev_photo.perceptual_hash)\n\n if distance <= similarity_threshold:\n current_group.append(photo)\n else:\n if len(current_group) >= 2:\n groups.append(current_group)\n current_group = [photo]\n\n if len(current_group) >= 2:\n groups.append(current_group)\n\n return groups\n\n\n# Default rules configuration\n# Hard criteria rules (enabled by default)\nDEFAULT_HARD_RULES = [\n {\n \"id\": 1,\n \"name\": \"EXIF Burst Mode Tag\",\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": True,\n \"is_default\": True,\n \"description\": \"Detects photos where camera was in burst mode (using MakerNotes:BurstMode or MakerNotes:ContinuousDrive EXIF tags)\",\n },\n {\n \"id\": 2,\n \"name\": \"EXIF Sequence Number\",\n \"rule_type\": BurstRuleTypes.EXIF_SEQUENCE_NUMBER,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": True,\n \"is_default\": True,\n \"description\": \"Groups photos by EXIF sequence number (MakerNotes:SequenceNumber) taken at the same time\",\n },\n {\n \"id\": 3,\n \"name\": \"Filename Burst Pattern\",\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": True,\n \"is_default\": True,\n \"pattern_type\": \"all\",\n \"description\": \"Detects burst sequences from filename patterns (e.g., IMG_001_BURST001, photo (1), photo (2))\",\n },\n]\n\n# Soft criteria rules (disabled by default)\nDEFAULT_SOFT_RULES = [\n {\n \"id\": 101,\n \"name\": \"Timestamp Proximity\",\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": False,\n \"is_default\": True,\n \"interval_ms\": 2000,\n \"require_same_camera\": True,\n \"description\": \"Groups photos taken within a short time interval (configurable, default 2 seconds)\",\n },\n {\n \"id\": 102,\n \"name\": \"Visual Similarity\",\n \"rule_type\": BurstRuleTypes.VISUAL_SIMILARITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": False,\n \"is_default\": True,\n \"similarity_threshold\": 15,\n \"description\": \"Groups visually similar consecutive photos using perceptual hash comparison\",\n },\n]\n\n# Other available rules (not included by default)\nOTHER_RULES = [\n {\n \"id\": 4,\n \"name\": \"Filename Burst Suffix Only\",\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": False,\n \"is_default\": False,\n \"pattern_type\": \"burst_suffix\",\n \"description\": \"Only detect files with explicit _BURST suffix in filename\",\n },\n {\n \"id\": 5,\n \"name\": \"Custom Filename Pattern\",\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": False,\n \"is_default\": False,\n \"pattern_type\": \"custom\",\n \"custom_pattern\": \"\",\n \"description\": \"Use a custom regex pattern to match burst filenames\",\n },\n {\n \"id\": 103,\n \"name\": \"Timestamp Proximity (Loose)\",\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": False,\n \"is_default\": False,\n \"interval_ms\": 5000,\n \"require_same_camera\": False,\n \"description\": \"Groups photos taken within 5 seconds, regardless of camera\",\n },\n]\n\n\ndef get_default_burst_detection_rules():\n \"\"\"Get default burst detection rules as JSON-serializable list.\"\"\"\n return DEFAULT_HARD_RULES + DEFAULT_SOFT_RULES\n\n\ndef get_all_predefined_burst_rules():\n \"\"\"Get all predefined burst detection rules (default + optional).\"\"\"\n return DEFAULT_HARD_RULES + DEFAULT_SOFT_RULES + OTHER_RULES\n\n\ndef _as_json(configs):\n \"\"\"Convert rule configs to JSON string.\"\"\"\n return json.dumps(configs, default=lambda x: x.__dict__)\n\n\n# Pre-computed JSON strings for API responses\nDEFAULT_RULES_JSON = _as_json(get_default_burst_detection_rules())\nPREDEFINED_RULES_JSON = _as_json(get_all_predefined_burst_rules())\n\n\ndef as_rules(configs):\n \"\"\"Convert list of rule configs to list of BurstDetectionRule objects.\"\"\"\n return [BurstDetectionRule(config) for config in configs]\n\n\ndef get_hard_rules(rules):\n \"\"\"Filter to only hard criteria rules.\"\"\"\n return [r for r in rules if r.category == BurstRuleCategory.HARD and r.enabled]\n\n\ndef get_soft_rules(rules):\n \"\"\"Filter to only soft criteria rules.\"\"\"\n return [r for r in rules if r.category == BurstRuleCategory.SOFT and r.enabled]\n\n\ndef get_enabled_rules(rules):\n \"\"\"Filter to only enabled rules.\"\"\"\n return [r for r in rules if r.enabled]\n" }, { "path": "api/cluster_manager.py", "content": "import numpy as np\n\nfrom api.models.cluster import UNKNOWN_CLUSTER_ID, Cluster, get_unknown_cluster\nfrom api.models.face import Face\nfrom api.models.person import Person, get_or_create_person\nfrom api.models.user import User\nfrom api.util import logger\n\n\nclass ClusterManager:\n @staticmethod\n def try_add_cluster(\n user: User, cluster_id: int, faces: list[Face], padLen: int = 1\n ) -> list[Cluster]:\n added_clusters: list[Cluster] = []\n known_faces: list[Face] = []\n face_ids_by_cluster: dict[int, list[int]] = dict()\n unknown_faces: list[Face] = []\n unknown_ids: list[int] = []\n encoding_by_person: dict[int, list[np.ndarray]] = dict()\n\n face: Face\n new_cluster: Cluster\n unknown_cluster: Cluster = get_unknown_cluster(user=user)\n labelStr = str(cluster_id).zfill(padLen)\n for face in faces:\n if not face.person:\n unknown_faces.append(face)\n unknown_ids.append(face.id)\n else:\n known_faces.append(face)\n\n if cluster_id == UNKNOWN_CLUSTER_ID:\n logger.info(\"Adding unknown cluster\")\n logger.info(\n \"Adding unknown %d faces to unknown cluster\" % len(unknown_faces)\n )\n logger.info(\"Adding known %d faces to unknown cluster\" % len(known_faces))\n for face in unknown_faces:\n face.cluster = unknown_cluster\n face.cluster_person = None\n face.save()\n for face in known_faces:\n face.cluster = unknown_cluster\n face.save()\n\n return added_clusters\n\n if len(known_faces) == 0:\n new_cluster: Cluster\n new_person: Person\n\n new_person = get_or_create_person(\n name=\"Unknown \" + labelStr, owner=user, kind=Person.KIND_CLUSTER\n )\n new_person.cluster_owner = user\n new_person.save()\n new_cluster = Cluster.get_or_create_cluster_by_id(user, cluster_id)\n new_cluster.name = \"Cluster \" + str(cluster_id)\n\n new_cluster.person = new_person\n encoding_by_person[new_cluster.person.id] = []\n new_cluster.save()\n added_clusters.append(new_cluster)\n\n for face in unknown_faces:\n encoding_by_person[new_cluster.person.id].append(\n face.get_encoding_array()\n )\n Face.objects.filter(id__in=unknown_ids).update(\n cluster=new_cluster,\n cluster_person=new_person,\n )\n else:\n clusters_by_person: dict[int, Cluster] = dict()\n mean_encoding_by_cluster: dict[int, list[np.ndarray]] = dict()\n idx: int = 0\n for face in known_faces:\n if face.person.id not in clusters_by_person.keys():\n idx = idx + 1\n new_cluster = Cluster.get_or_create_cluster_by_name(\n user, \"Cluster \" + str(cluster_id) + \"-\" + str(idx)\n )\n new_cluster.cluster_id = cluster_id\n new_cluster.person = face.person\n clusters_by_person[new_cluster.person.id] = new_cluster\n added_clusters.append(new_cluster)\n encoding_by_person[face.person.id] = []\n face_ids_by_cluster[new_cluster.id] = []\n else:\n new_cluster = clusters_by_person[face.person.id]\n encoding_by_person[face.person.id].append(face.get_encoding_array())\n face_ids_by_cluster[new_cluster.id].append(face.id)\n for new_cluster in added_clusters:\n Face.objects.filter(id__in=face_ids_by_cluster[new_cluster.id]).update(\n cluster=new_cluster\n )\n\n # Set initial metadata on the split clusters based on known faces\n for new_cluster in added_clusters:\n new_cluster.set_metadata(encoding_by_person[new_cluster.person.id])\n mean_encoding_by_cluster[new_cluster.id] = (\n new_cluster.get_mean_encoding_array()\n )\n\n # Clear the face IDs list to prepare for processing the unknown faces\n for new_cluster in added_clusters:\n face_ids_by_cluster[new_cluster.id] = []\n\n for new_cluster in added_clusters:\n Face.objects.filter(id__in=face_ids_by_cluster[new_cluster.id]).update(\n cluster=new_cluster,\n cluster_person=new_cluster.person,\n )\n\n # Update statistics again and save everything, since we've added more faces\n for new_cluster in added_clusters:\n new_cluster.set_metadata(encoding_by_person[new_cluster.person.id])\n new_cluster.save()\n\n return added_clusters\n" }, { "path": "api/date_time_extractor.py", "content": "import json\nimport math\nimport os\nimport pathlib\nimport re\nfrom datetime import datetime\n\nimport pytz\n\nfrom api.metadata.tags import Tags\nfrom api.util import logger\n\n\ndef _regexp_group_range(a, b):\n return \"(\" + \"|\".join(f\"{i:02}\" for i in range(a, b)) + \")\"\n\n\n_REGEXP_GROUP_YEAR = r\"((?:19|20|21)\\d\\d)\"\n_REGEXP_GROUP_MONTH = _regexp_group_range(1, 13)\n_REGEXP_GROUP_DAY = _regexp_group_range(1, 32)\n_REGEXP_GROUP_HOUR = _regexp_group_range(0, 24)\n_REGEXP_GROUP_MIN = r\"([0-5]\\d)\"\n_REGEXP_GROUP_SEC = r\"([0-5]\\d)\"\n_REGEXP_DELIM = r\"[-:_\\., ]*\"\n_NOT_A_NUMBER = r\"(? len(group_mapping):\n raise ValueError(\n f\"Can't have more groups than group mapping values: {x}, regexp: {regexp}, mapping: {group_mapping}\"\n )\n datetime_args = [\n None,\n None,\n None,\n 0,\n 0,\n 0,\n 0,\n ] # year, month, day, hour, minute, second, microsecond\n for value, how_to_use in zip(g, group_mapping):\n if how_to_use not in REGEXP_GROUP_MAPPINGS:\n raise ValueError(\n f\"Group mapping {how_to_use} is unknown - must be one of {list(REGEXP_GROUP_MAPPINGS.keys())}\"\n )\n ind = REGEXP_GROUP_MAPPINGS[how_to_use]\n # handle case when we have less groups than expected\n if value is not None:\n datetime_args[ind] = int(value)\n\n try:\n parsed_datetime = datetime(*datetime_args)\n delta = parsed_datetime - datetime.now()\n if delta.days > 30:\n logger.error(\n f\"Error while parsing datetime from '{x}': Parsed datetime is {delta.days} in the future.\"\n )\n return None\n\n return parsed_datetime\n except ValueError:\n logger.error(\n f\"Error while trying to create datetime using '{x}': datetime arguments {datetime_args}. Regexp used: '{regexp}'\"\n )\n return None\n\n\nclass RuleTypes:\n EXIF = \"exif\"\n PATH = \"path\"\n FILESYSTEM = \"filesystem\"\n USER_DEFINED = \"user_defined\"\n\n\nclass TimeExtractionRule:\n \"\"\"The goal is to help extract local time, but for historical reason it is expected the returned\n datetime will have timezone to be set to pytz.utc (so local time + timezone equal to UTC)..\n\n Some sources of data might give us very rich information, e.g. timestamp + timezone,\n but others only allow to get local time (without knowing real timestamp).\n\n The logic for extracting local time is described as a list of rules that should be applied\n one after another until one rule is able to extract date time (or until all rules are tried\n without success).\n\n Currently supported rule types:\n - \"exif\" - local time is taken using exif tag params[\"exif_tag\"] as obtained with exiftool\n - \"path\" - time is taken from the filename using a regular expression matching\n - if params[\"path_part\"] is set to \"full_path\" then full path (as seen by backend container)\n is used for regexp matching instead of just fileanme.\n - if params[\"custom_regexp\"] is specified - that regexp is used instead of default one\n (it is still expecting 6 groups to be matched: year, month, day, hour, minute, second).\n - \"fs\" - time is taken from file property. Since these are unix timestamps without timezones\n they are always translated to local time using UTC.\n - params[\"file_property\"] must be specified and equal to one of the following:\n - \"mtime\" - for file modifided time\n - \"ctime\" - for file created time\n If a rule can't fetch the time (e.g. the exif tag value is not present or path doesn't match\n a regexp) then that rule is considered to be not applicable.\n\n In some cases it is known that the local time the rule would obtain is not in the desired\n timezone. E.g. video datetime tag QuickTime:CreateDate is by standard written in UTC rather\n than local time. For such cases each rule can optionally have setting \"transform_tz\" set to \"1\"\n - in that case this rule should also specify \"source_tz\" and \"report_tz\" settings where\n \"source_tz\" is describing the timezone that the rule is getting and \"report_tz\" is describing\n the timezone of the location where the photo/video was taken. Both \"source_tz\" and \"report_tz\"\n should be one of the following:\n - \"utc\" - UTC timezone\n - \"gps_timezonefinder\" - the timezone of the GPS location associated with the photo/video\n - \"server_local\" - the timezone of the librephotos server - not very useful since we run docker containers in UTC timezone.\n - \"user_default\" - user default timezone\n - \"name:\" - the timezone with the name \n If either \"source_tz\" or \"report_tz\" could not be obtained the rule is considered to be not applicable.\n\n Additionally each rule can have condition specifications that limits the rule application\n to only the photos/videos that meet the condition's requirement. Supported conditions:\n - \"condition_path\": \"\" - rule only applied to files with full path (as seen by backend)\n matching the regexp\n - \"condition_filename\": \"\" - rule only applied to files with filename matching the regexp\n - \"condition_exif\": \"//\" - first \"//\" is considered end of tag name and the rule is only\n applied if value of tag exists and matches the regexp.\n\n If multiple conditions are provided the rule is considered applicable only if all of them are met.\n\n Examples of the rules:\n - Take local time from exif tag \"EXIF:DateTimeOriginal\" if it is available:\n\n {\n \"rule_type\": \"exif\",\n \"exif_tag\": \"EXIF:DateTimeOriginal\",\n }\n\n - Take UTC time using tag \"QuickTime:CreateDate\" and convert it from UTC\n to timezone associated with the GPS coordinates (only applies if both\n tag value and GPS coordinates are available):\n\n {\n \"rule_type\": \"exif\",\n \"exif_tag\": \"QuickTime:CreateDate\",\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"gps_timezonefinder\",\n }\n\n - Look at the filename and try to extract time from it - treat it as local time\n (it is known that some devices are not using local time here - e.g. some phones\n might use UTC time for video filenames):\n\n {\n \"rule_type\": \"path\",\n }\n\n - Take UTC time time using tag \"QuickTime:CreateDate\" and convert it from UTC\n to a fixed timezone \"Europe/Moscow\" to get local time. Only apply to files\n which path contains \"Moscow_Visit\" or \"FromMoscow\":\n\n {\n \"rule_type\": \"exif\",\n \"exif_tag\": \"QuickTime:CreateDate\",\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"name:Europe/Moscow\",\n \"condition_path\": \"(Moscow_Visit|FromMoscow)\",\n }\n\n - Take modified time of the file and get local time using timezone associated\n with the GPS location. Only apply to files with \"EXIF:Model\" exif tag\n containing \"FooBar\":\n\n {\n \"rule_type\": \"filesystem\",\n \"file_property\": \"mtime\",\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"gps_timezonefinder\",\n \"condition_exif\": \"EXIF:Model//FooBar\"\n }\n \"\"\"\n\n def __init__(self, params):\n self.rule_type = params[\"rule_type\"]\n self.params = params\n\n def get_required_exif_tags(self):\n condition_tag, pattern = self._get_condition_exif()\n res = set()\n if condition_tag is not None:\n res.add(condition_tag)\n if self.rule_type == RuleTypes.EXIF:\n res.add(self.params[\"exif_tag\"])\n return res\n\n def _get_no_tz_dt_from_tag(self, tag_name, exif_tags):\n tag_val = exif_tags.get(tag_name)\n if not tag_val:\n return None\n dt = _extract_no_tz_datetime_from_str(tag_val)\n return dt\n\n def _check_condition_path(self, path):\n if \"condition_path\" in self.params:\n return re.search(self.params[\"condition_path\"], path) is not None\n else:\n return True\n\n def _check_condition_filename(self, path):\n if \"condition_filename\" in self.params:\n return (\n re.search(self.params[\"condition_filename\"], pathlib.Path(path).name)\n is not None\n )\n else:\n return True\n\n def _get_condition_exif(self):\n val = self.params.get(\"condition_exif\")\n if val is None:\n return None, None\n tag_and_pattern = val.split(\"//\", maxsplit=1)\n if len(tag_and_pattern) != 2:\n raise ValueError(\n f\"Value of condition_exif must contain '//' delimiter between tag name and pattern: '{val}'\"\n )\n tag, pattern = tag_and_pattern\n return tag, pattern\n\n def _check_condition_exif(self, exif_tags):\n tag, pattern = self._get_condition_exif()\n if tag:\n tag_value = exif_tags.get(tag)\n if not tag_value:\n return False\n return re.search(pattern, tag_value) is not None\n else:\n return True\n\n def _check_conditions(self, path, exif_tags, gps_lat, gps_lon):\n return (\n self._check_condition_exif(exif_tags)\n and self._check_condition_path(path)\n and self._check_condition_filename(path)\n )\n\n def apply(\n self, path, exif_tags, gps_lat, gps_lon, user_default_tz, user_defined_timestamp\n ):\n if not self._check_conditions(path, exif_tags, gps_lat, gps_lon):\n return None\n if self.rule_type == RuleTypes.EXIF:\n return self._apply_exif(exif_tags, gps_lat, gps_lon, user_default_tz)\n elif self.rule_type == RuleTypes.PATH:\n return self._apply_path(path, gps_lat, gps_lon, user_default_tz)\n elif self.rule_type == RuleTypes.FILESYSTEM:\n return self._apply_filesystem(path, gps_lat, gps_lon, user_default_tz)\n elif self.rule_type == RuleTypes.USER_DEFINED:\n return user_defined_timestamp\n else:\n raise ValueError(f\"Unknown rule type {self.rule_type}\")\n\n def _get_tz(self, description, gps_lat, gps_lon, user_default_tz):\n \"\"\"None is a valid timezone returned here (meaning that we want to use server local time).\n This is why this function returns a tuple with the first element specifying success of\n determining the timezone, and the second element - the timezone itself.\n \"\"\"\n if description == \"gps_timezonefinder\":\n if not _check_gps_ok(gps_lat, gps_lon):\n return (False, None)\n from timezonefinder import TimezoneFinder\n\n tzfinder = TimezoneFinder()\n tz_name = tzfinder.timezone_at(lng=gps_lon, lat=gps_lat)\n return (True, pytz.timezone(tz_name)) if tz_name else (False, None)\n elif description == \"user_default\":\n return (True, pytz.timezone(user_default_tz))\n elif description == \"server_local\":\n return (True, None)\n elif description.lower() == \"utc\":\n return (True, pytz.utc)\n elif description.startswith(\"name:\"):\n return (True, pytz.timezone(description[5:]))\n else:\n raise ValueError(f\"Unknown tz description {description}\")\n\n def _transform_tz(self, dt, gps_lat, gps_lon, user_default_tz):\n if not dt:\n return None\n if self.params.get(\"transform_tz\"):\n has_source_tz, source_tz = self._get_tz(\n self.params[\"source_tz\"], gps_lat, gps_lon, user_default_tz\n )\n if not has_source_tz:\n return None\n has_report_tz, report_tz = self._get_tz(\n self.params[\"report_tz\"], gps_lat, gps_lon, user_default_tz\n )\n if not has_report_tz:\n return None\n # Either of source_tz or report_tz might be None - meaning that we want to use\n # server local timezone\n dt = datetime.fromtimestamp(\n dt.replace(tzinfo=source_tz).timestamp(), report_tz\n )\n return dt.replace(tzinfo=pytz.utc)\n\n def _apply_exif(self, exif_tags, gps_lat, gps_lon, user_default_tz):\n dt = self._get_no_tz_dt_from_tag(self.params[\"exif_tag\"], exif_tags)\n return self._transform_tz(dt, gps_lat, gps_lon, user_default_tz)\n\n def _apply_path(self, path, gps_lat, gps_lon, user_default_tz):\n path_part = self.params.get(\"path_part\")\n if path_part is None or path_part == \"filename\":\n source = pathlib.Path(path).name\n elif path_part == \"full_path\":\n source = path\n else:\n raise ValueError(f\"Unknown path_part {path_part}\")\n\n group_mapping = None\n regexp = self.params.get(\"custom_regexp\")\n if not regexp:\n predefined_regexp_type = self.params.get(\"predefined_regexp\", \"default\")\n if predefined_regexp_type not in PREDEFINED_REGEXPS:\n raise ValueError(\n f\"Unknown predefined regexp type {predefined_regexp_type}\"\n )\n regexp, group_mapping = PREDEFINED_REGEXPS[predefined_regexp_type]\n dt = _extract_no_tz_datetime_from_str(source, regexp, group_mapping)\n return self._transform_tz(dt, gps_lat, gps_lon, user_default_tz)\n\n def _apply_filesystem(self, path, gps_lat, gps_lon, user_default_tz):\n file_property = self.params.get(\"file_property\")\n if file_property == \"mtime\":\n dt = datetime.fromtimestamp(os.path.getmtime(path), pytz.utc)\n elif file_property == \"ctime\":\n dt = datetime.fromtimestamp(os.path.getctime(path), pytz.utc)\n else:\n raise ValueError(f\"Unknown file_property {file_property}\")\n return self._transform_tz(dt, gps_lat, gps_lon, user_default_tz)\n\n\ndef _check_gps_ok(lat, lon):\n return (\n lat is not None\n and lon is not None\n and math.isfinite(lat)\n and math.isfinite(lon)\n and (lat != 0.0 or lon != 0.0)\n )\n\n\nALL_TIME_ZONES = pytz.all_timezones\n\nDEFAULT_RULES_PARAMS = [\n {\n \"id\": 14,\n \"name\": \"Timestamp set by user\",\n \"rule_type\": RuleTypes.USER_DEFINED,\n },\n {\n \"id\": 15,\n \"name\": f\"Local time from {Tags.DATE_TIME} exif tag\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.DATE_TIME,\n },\n {\n \"id\": 1,\n \"name\": f\"Local time from {Tags.DATE_TIME_ORIGINAL} exif tag\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.DATE_TIME_ORIGINAL,\n },\n {\n \"id\": 2,\n \"name\": \"Get Video creation tag in UTC + figure out timezone using GPS coordinates\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.QUICKTIME_CREATE_DATE,\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"gps_timezonefinder\",\n },\n {\n \"id\": 11,\n \"name\": f\"Use {Tags.GPS_DATE_TIME} tag + figure out timezone using GPS coordinates\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.GPS_DATE_TIME,\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"gps_timezonefinder\",\n },\n {\n \"id\": 3,\n \"name\": \"Using filename assuming time is local (most of filenames auto generated by smartphones etc)\",\n \"rule_type\": RuleTypes.PATH,\n },\n {\n \"id\": 4,\n \"name\": \"Video creation datetime in user default timezone (can't find out actual timezone)\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.QUICKTIME_CREATE_DATE,\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"user_default\",\n },\n {\n \"id\": 5,\n \"name\": \"Extract date using WhatsApp file name\",\n \"rule_type\": RuleTypes.PATH,\n \"predefined_regexp\": \"whatsapp\",\n },\n]\n\nOTHER_RULES_PARAMS = [\n {\n \"id\": 6,\n \"name\": \"Video creation datetime in UTC timezone (can't find out actual timezone)\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.QUICKTIME_CREATE_DATE,\n },\n {\n \"id\": 7,\n \"name\": \"File modified time in user default timezone\",\n \"rule_type\": RuleTypes.FILESYSTEM,\n \"file_property\": \"mtime\",\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"user_default\",\n },\n {\n \"id\": 8,\n \"name\": \"File modified time in UTC timezone\",\n \"rule_type\": RuleTypes.FILESYSTEM,\n \"file_property\": \"mtime\",\n },\n {\n \"id\": 9,\n \"name\": \"File created time in user default timezone\",\n \"rule_type\": RuleTypes.FILESYSTEM,\n \"file_property\": \"ctime\",\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"user_default\",\n },\n {\n \"id\": 10,\n \"name\": \"File created time in UTC timezone\",\n \"rule_type\": RuleTypes.FILESYSTEM,\n \"file_property\": \"ctime\",\n },\n {\n \"id\": 12,\n \"name\": f\"Use {Tags.GPS_DATE_TIME} tag in user default timezone (can't find out actual timezone)\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.GPS_DATE_TIME,\n \"transform_tz\": 1,\n \"source_tz\": \"utc\",\n \"report_tz\": \"user_default\",\n },\n {\n \"id\": 13,\n \"name\": f\"Use {Tags.GPS_DATE_TIME} tag in UTC timezone (can't find out actual timezone)\",\n \"rule_type\": RuleTypes.EXIF,\n \"exif_tag\": Tags.GPS_DATE_TIME,\n },\n]\n\n\ndef set_as_default_rule(rule):\n rule[\"is_default\"] = True\n return rule\n\n\ndef set_as_other_rule(rule):\n rule[\"is_default\"] = False\n return rule\n\n\nPREDEFINED_RULES_PARAMS = list(map(set_as_default_rule, DEFAULT_RULES_PARAMS)) + list(\n map(set_as_other_rule, OTHER_RULES_PARAMS)\n)\n\n\ndef _as_json(configs):\n return json.dumps(configs, default=lambda x: x.__dict__)\n\n\nDEFAULT_RULES_JSON = _as_json(DEFAULT_RULES_PARAMS)\nPREDEFINED_RULES_JSON = _as_json(PREDEFINED_RULES_PARAMS)\nALL_TIME_ZONES_JSON = _as_json(ALL_TIME_ZONES)\n\n\ndef as_rules(configs):\n return list(map(TimeExtractionRule, configs))\n\n\ndef extract_local_date_time(\n path, rules, exif_getter, gps_lat, gps_lon, user_default_tz, user_defined_timestamp\n):\n required_tags = set()\n for rule in rules:\n required_tags.update(rule.get_required_exif_tags())\n required_tags = list(required_tags)\n exif_values = exif_getter(required_tags)\n exif_tags = {k: v for k, v in zip(required_tags, exif_values)}\n for rule in rules:\n res = rule.apply(\n path, exif_tags, gps_lat, gps_lon, user_default_tz, user_defined_timestamp\n )\n if res:\n return res\n return None\n" }, { "path": "api/directory_watcher/__init__.py", "content": "\"\"\"\nDirectory watcher module for scanning and processing photos.\n\nThis module implements a two-phase scan architecture to avoid race conditions\nwhen processing RAW+JPEG pairs concurrently:\n\nPhase 1: Collect all files and group by (directory, basename)\n - IMG_001.jpg, IMG_001.CR2, IMG_001.xmp -> one group\n - IMG_002.jpg -> separate group\n\nPhase 2: Process each group, creating one Photo per group\n with all file variants attached.\n\"\"\"\n\n# Main scan functions\nfrom api.directory_watcher.scan_jobs import (\n scan_photos,\n scan_missing_photos,\n photo_scanner,\n)\n\n# File handling\nfrom api.directory_watcher.file_handlers import (\n create_new_image,\n create_file_record,\n group_files_into_photo,\n handle_new_image,\n handle_file_group,\n)\n\n# File grouping utilities\nfrom api.directory_watcher.file_grouping import (\n JPEG_EXTENSIONS,\n FILE_TYPE_PRIORITY,\n get_file_grouping_key,\n select_main_file,\n find_matching_jpeg_photo,\n find_matching_image_for_video,\n)\n\n# Processing jobs\nfrom api.directory_watcher.processing_jobs import (\n generate_tags,\n generate_tag_job,\n add_geolocation,\n geolocation_job,\n scan_faces,\n generate_face_embeddings,\n)\n\n# Repair jobs\nfrom api.directory_watcher.repair_jobs import (\n repair_ungrouped_file_variants,\n)\n\n# Utilities\nfrom api.directory_watcher.utils import (\n is_hidden,\n should_skip,\n walk_directory,\n walk_files,\n update_scan_counter,\n)\n\n# Re-export from api.models.file for backwards compatibility\nfrom api.models.file import is_valid_media\n\n__all__ = [\n # Scan jobs\n \"scan_photos\",\n \"scan_missing_photos\",\n \"photo_scanner\",\n # File handling\n \"create_new_image\",\n \"create_file_record\",\n \"group_files_into_photo\",\n \"handle_new_image\",\n \"handle_file_group\",\n # File grouping\n \"JPEG_EXTENSIONS\",\n \"FILE_TYPE_PRIORITY\",\n \"get_file_grouping_key\",\n \"select_main_file\",\n \"find_matching_jpeg_photo\",\n \"find_matching_image_for_video\",\n # Processing jobs\n \"generate_tags\",\n \"generate_tag_job\",\n \"add_geolocation\",\n \"geolocation_job\",\n \"scan_faces\",\n \"generate_face_embeddings\",\n # Repair jobs\n \"repair_ungrouped_file_variants\",\n # Utilities\n \"is_hidden\",\n \"should_skip\",\n \"walk_directory\",\n \"walk_files\",\n \"update_scan_counter\",\n # Re-exported from api.models.file\n \"is_valid_media\",\n]\n" }, { "path": "api/directory_watcher/file_grouping.py", "content": "\"\"\"\nFile grouping utilities for the two-phase scan architecture.\n\nThis module provides functions for grouping related files (RAW+JPEG pairs,\nLive Photos, etc.) so they can be processed together as a single Photo.\n\"\"\"\n\nimport os\n\nfrom api.models import File, Photo\n\n\n# JPEG/image extensions that RAW files can be paired with\nJPEG_EXTENSIONS = {'.jpg', '.jpeg', '.heic', '.heif', '.png', '.tiff', '.tif'}\n\n# File type priority for main_file selection (lower number = higher priority)\n# JPEG/processed images should be main_file, RAW/video variants are secondary\nFILE_TYPE_PRIORITY = {\n File.IMAGE: 1, # JPEG, HEIC, PNG - highest priority\n File.VIDEO: 2, # Videos (standalone or Live Photo motion)\n File.RAW_FILE: 3, # RAW files are variants, not main\n File.METADATA_FILE: 4, # XMP sidecars - lowest priority\n File.UNKNOWN: 5,\n}\n\n\ndef get_file_grouping_key(path: str) -> tuple[str, str]:\n \"\"\"\n Get the grouping key for a file path.\n \n Files with the same (directory, basename) should be grouped together\n as variants of the same Photo (e.g., IMG_001.jpg + IMG_001.CR2 + IMG_001.xmp).\n \n Args:\n path: File path to get grouping key for\n \n Returns:\n Tuple of (directory, lowercase_basename_without_extension)\n \"\"\"\n directory = os.path.dirname(path)\n basename = os.path.splitext(os.path.basename(path))[0].lower()\n return (directory, basename)\n\n\ndef select_main_file(files: list[File]) -> File | None:\n \"\"\"\n Select the best file to be the main_file for a Photo.\n \n Priority: IMAGE > VIDEO > RAW > METADATA\n Within same type, prefer the first one found (alphabetically by path).\n \n Args:\n files: List of File objects to choose from\n \n Returns:\n The File that should be main_file, or None if empty list\n \"\"\"\n if not files:\n return None\n \n return min(files, key=lambda f: (FILE_TYPE_PRIORITY.get(f.type, 999), f.path))\n\n\ndef find_matching_jpeg_photo(raw_path: str, user) -> Photo | None:\n \"\"\"\n Find an existing Photo with a matching JPEG/image file for a RAW file.\n \n Matches based on same base filename (without extension) in the same directory.\n This implements the PhotoPrism-like file variant model where RAW+JPEG are\n one Photo with multiple file variants, not separate Photos.\n \n Args:\n raw_path: Path to the RAW file\n user: Owner of the photos\n \n Returns:\n Matching Photo if found, None otherwise\n \"\"\"\n raw_dir = os.path.dirname(raw_path)\n raw_basename = os.path.splitext(os.path.basename(raw_path))[0]\n \n # Look for matching JPEG/image file in same directory\n for jpeg_ext in JPEG_EXTENSIONS:\n # Try both lowercase and uppercase extensions\n for ext in [jpeg_ext, jpeg_ext.upper()]:\n jpeg_path = os.path.join(raw_dir, raw_basename + ext)\n photo = Photo.objects.filter(\n owner=user,\n main_file__path=jpeg_path\n ).first()\n if photo:\n return photo\n \n return None\n\n\ndef find_matching_image_for_video(video_path: str, user) -> Photo | None:\n \"\"\"\n Find an existing Photo with a matching image file for a Live Photo video.\n \n Apple Live Photos store the video as a separate .mov file with the same\n base name as the image. This allows attaching the video as a file variant.\n \n Args:\n video_path: Path to the video file\n user: Owner of the photos\n \n Returns:\n Matching Photo if found, None otherwise\n \"\"\"\n video_dir = os.path.dirname(video_path)\n video_basename = os.path.splitext(os.path.basename(video_path))[0]\n video_ext_lower = os.path.splitext(video_path)[1].lower()\n \n # Only match .mov files (Apple Live Photos) with same base name\n if video_ext_lower not in ['.mov']:\n return None\n \n # Look for matching image file with same base name\n image_extensions = list(JPEG_EXTENSIONS) + ['.heic', '.HEIC']\n for img_ext in image_extensions:\n for ext in [img_ext, img_ext.upper()]:\n image_path = os.path.join(video_dir, video_basename + ext)\n photo = Photo.objects.filter(\n owner=user,\n main_file__path=image_path\n ).first()\n if photo:\n return photo\n \n return None\n" }, { "path": "api/directory_watcher/file_handlers.py", "content": "\"\"\"\nFile and Photo creation handlers.\n\nThis module contains functions for creating File records and grouping\nthem into Photo objects.\n\"\"\"\n\nimport datetime\nimport os\n\nimport pytz\nfrom django.conf import settings\nfrom django.db.models import Q\n\nfrom api import util\nfrom api.models import File, Photo, Thumbnail\nfrom api.models.file import calculate_hash, is_metadata, is_raw, is_valid_media, is_video\nfrom api.models.photo_search import PhotoSearch\nfrom api.perceptual_hash import calculate_hash_from_thumbnail\nfrom api.stacks.live_photo import has_embedded_motion_video, extract_embedded_motion_video\n\nfrom api.directory_watcher.file_grouping import (\n FILE_TYPE_PRIORITY,\n find_matching_jpeg_photo,\n find_matching_image_for_video,\n select_main_file,\n)\nfrom api.directory_watcher.utils import update_scan_counter\n\n\ndef create_file_record(user, path) -> File | None:\n \"\"\"\n Phase 1: Create a File record for a path without creating/grouping Photos.\n \n This is the first phase of the two-phase scan architecture:\n - Phase 1: Create File records for all discovered files (this function)\n - Phase 2: Group files into Photos by (directory, basename)\n \n This separation eliminates race conditions where concurrent processing\n of RAW and JPEG files could create separate Photos instead of grouping them.\n \n Args:\n user: The owner of the file\n path: The file path\n \n Returns:\n File object if created/found, None if invalid media\n \"\"\"\n if not is_valid_media(path=path, user=user):\n return None\n \n hash_value = calculate_hash(user, path)\n \n # Skip if this is embedded media (already attached to another file)\n if File.embedded_media.through.objects.filter(Q(to_file_id=hash_value)).exists():\n util.logger.warning(f\"embedded content file found {path}\")\n return None\n \n # Create the File record (File.create handles race conditions via unique path constraint)\n file = File.create(path, user)\n return file\n\n\ndef group_files_into_photo(user, files: list[File], job_id) -> Photo | None:\n \"\"\"\n Phase 2: Group a list of related files into a single Photo.\n \n Creates a new Photo with the given files as variants, selecting the\n best file as main_file based on type priority (IMAGE > VIDEO > RAW > METADATA).\n \n This function should be called with all files that share the same\n (directory, basename) - e.g., IMG_001.jpg, IMG_001.CR2, IMG_001.xmp.\n \n Args:\n user: The owner of the photo\n files: List of File objects to group (must not be empty)\n job_id: Job ID for logging\n \n Returns:\n The created Photo, or None if no valid files\n \"\"\"\n if not files:\n return None\n \n # Filter out metadata files for main photo creation - they're sidecars\n non_metadata_files = [f for f in files if f.type != File.METADATA_FILE]\n \n if not non_metadata_files:\n # Only metadata files - no photo to create\n util.logger.warning(f\"job {job_id}: Only metadata files in group, skipping\")\n return None\n \n # Select main file based on priority\n main_file = select_main_file(non_metadata_files)\n if not main_file:\n return None\n \n # Check if a Photo already exists with any of these files\n existing_photo = Photo.objects.filter(\n owner=user,\n files__in=files\n ).first()\n \n if existing_photo:\n # Add any missing files to the existing photo\n for f in files:\n if not existing_photo.files.filter(hash=f.hash).exists():\n existing_photo.files.add(f)\n util.logger.info(f\"job {job_id}: Attached file {f.path} to existing Photo {existing_photo.image_hash}\")\n \n # Update main_file if current one has lower priority\n if existing_photo.main_file:\n current_priority = FILE_TYPE_PRIORITY.get(existing_photo.main_file.type, 999)\n new_priority = FILE_TYPE_PRIORITY.get(main_file.type, 999)\n if new_priority < current_priority:\n existing_photo.main_file = main_file\n existing_photo.save(update_fields=['main_file'])\n \n return existing_photo\n \n # Create new Photo\n photo = Photo()\n photo.image_hash = main_file.hash\n photo.owner = user\n photo.added_on = datetime.datetime.now().replace(tzinfo=pytz.utc)\n photo.geolocation_json = {}\n photo.video = (main_file.type == File.VIDEO)\n photo.save()\n \n # Add all files to the photo\n for f in files:\n photo.files.add(f)\n \n photo.main_file = main_file\n photo.save()\n \n # Handle embedded media (Google/Samsung Live Photos with embedded video)\n if has_embedded_motion_video(main_file.path) and settings.FEATURE_PROCESS_EMBEDDED_MEDIA:\n em_path = extract_embedded_motion_video(main_file.path, main_file.hash)\n if em_path:\n em_file = File.create(em_path, user)\n main_file.embedded_media.add(em_file)\n photo.files.add(em_file)\n photo.save()\n \n util.logger.info(f\"job {job_id}: Created Photo {photo.image_hash} with {len(files)} file(s)\")\n return photo\n\n\ndef create_new_image(user, path) -> Photo | None:\n \"\"\"\n Creates a new Photo object based on user input and file path.\n \n This is the legacy single-file creation function, kept for backwards\n compatibility with upload handling. For scan operations, use the\n two-phase approach (create_file_record + group_files_into_photo).\n\n Args:\n user: The owner of the photo.\n path: The file path of the image.\n\n Returns:\n The created Photo object if successful, otherwise returns None.\n\n Note:\n This function implements file variant grouping (PhotoPrism-like):\n - RAW files are attached to existing JPEG Photos as file variants\n - Live Photo videos (.mov) are attached to existing image Photos as file variants\n - Other files create new Photo entities\n \"\"\"\n if not is_valid_media(path=path, user=user):\n return None\n hash_value = calculate_hash(user, path)\n if File.embedded_media.through.objects.filter(Q(to_file_id=hash_value)).exists():\n util.logger.warning(f\"embedded content file found {path}\")\n return None\n\n # Handle metadata files (XMP sidecars)\n if is_metadata(path):\n photo_name = os.path.splitext(os.path.basename(path))[0]\n photo_dir = os.path.dirname(path)\n photo = Photo.objects.filter(\n Q(files__path__contains=photo_dir)\n & Q(files__path__contains=photo_name)\n & ~Q(files__path__contains=os.path.basename(path))\n ).first()\n\n if photo:\n file = File.create(path, user)\n photo.files.add(file)\n photo.save()\n else:\n util.logger.warning(f\"no photo to metadata file found {path}\")\n return None\n\n # === File Variant Handling (PhotoPrism-like model) ===\n \n # Handle RAW files: attach to existing JPEG Photo if found\n if is_raw(path):\n existing_photo = find_matching_jpeg_photo(path, user)\n if existing_photo:\n # Check if this RAW file is already attached\n if not existing_photo.files.filter(path=path).exists():\n raw_file = File.create(path, user)\n existing_photo.files.add(raw_file)\n existing_photo.save()\n util.logger.info(f\"Attached RAW file {path} to existing Photo {existing_photo.image_hash}\")\n return existing_photo\n \n # Handle Live Photo videos (.mov): attach to existing image Photo if found\n if is_video(path):\n existing_photo = find_matching_image_for_video(path, user)\n if existing_photo:\n # Check if this video is already attached\n if not existing_photo.files.filter(path=path).exists():\n video_file = File.create(path, user)\n existing_photo.files.add(video_file)\n existing_photo.video = False # Keep photo as image (video is just a variant)\n existing_photo.save()\n util.logger.info(f\"Attached Live Photo video {path} to existing Photo {existing_photo.image_hash}\")\n return existing_photo\n\n # === Standard Photo Creation ===\n photo = Photo()\n photo.image_hash = hash_value\n photo.owner = user\n photo.added_on = datetime.datetime.now().replace(tzinfo=pytz.utc)\n photo.geolocation_json = {}\n photo.video = is_video(path)\n photo.save()\n file = File.create(path, user)\n \n # Live Photo detection - extracts embedded motion video if present (Google/Samsung)\n if has_embedded_motion_video(file.path) and settings.FEATURE_PROCESS_EMBEDDED_MEDIA:\n em_path = extract_embedded_motion_video(file.path, file.hash)\n if em_path:\n em_file = File.create(em_path, user)\n file.embedded_media.add(em_file)\n # Also add embedded video to Photo.files as a variant\n photo.files.add(em_file)\n \n photo.files.add(file)\n photo.main_file = file\n photo.save()\n return photo\n\n\ndef handle_new_image(user, path, job_id, photo=None):\n \"\"\"\n Handles the creation and all the processing of the photo needed for it to be displayed.\n\n Args:\n user: The owner of the photo.\n path: The file path of the image.\n job_id: The long-running job id, which gets updated when the task runs\n photo: An optional parameter, where you can input a photo instead of creating a new one. Used for uploading.\n\n Note:\n This function is used when uploading a picture, because rescanning does not perform machine learning tasks.\n \"\"\"\n try:\n start = datetime.datetime.now()\n if photo is None:\n photo = create_new_image(user, path)\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(f\"job {job_id}: save image: {path}, elapsed: {elapsed}\")\n if photo:\n _process_photo(photo, path, job_id, start)\n\n except Exception as e:\n try:\n util.logger.exception(\n f\"job {job_id}: could not load image {path}. reason: {str(e)}\"\n )\n except Exception:\n util.logger.exception(f\"job {job_id}: could not load image {path}\")\n finally:\n update_scan_counter(job_id)\n\n\ndef handle_file_group(user, file_paths: list[str], job_id):\n \"\"\"\n Phase 2 handler: Process a group of related files into a single Photo.\n \n This is called after Phase 1 has created File records for all paths.\n Files are grouped by (directory, basename) so RAW+JPEG pairs are processed together.\n \n Args:\n user: The owner of the files\n file_paths: List of file paths that share the same (directory, basename)\n job_id: Job ID for logging and progress tracking\n \"\"\"\n try:\n start = datetime.datetime.now()\n \n # Get or create File records for all paths\n files = []\n for path in file_paths:\n file = create_file_record(user, path)\n if file:\n files.append(file)\n \n if not files:\n util.logger.warning(f\"job {job_id}: No valid files in group: {file_paths}\")\n return\n \n # Group files into a Photo\n photo = group_files_into_photo(user, files, job_id)\n \n if not photo:\n util.logger.warning(f\"job {job_id}: Could not create photo for files: {file_paths}\")\n return\n \n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(f\"job {job_id}: created photo with {len(files)} files, elapsed: {elapsed}\")\n \n # Process the photo (thumbnails, EXIF, etc.) using main_file\n if photo.main_file:\n _process_photo(photo, photo.main_file.path, job_id, start)\n\n except Exception as e:\n try:\n util.logger.exception(\n f\"job {job_id}: could not process file group {file_paths}. reason: {str(e)}\"\n )\n except Exception:\n util.logger.exception(f\"job {job_id}: could not process file group\")\n finally:\n update_scan_counter(job_id)\n\n\ndef _process_photo(photo: Photo, path: str, job_id, start: datetime.datetime):\n \"\"\"\n Process a photo: generate thumbnails, extract EXIF, calculate hashes, etc.\n \n This is the common processing logic shared between handle_new_image and handle_file_group.\n \n Args:\n photo: The Photo object to process\n path: The main file path (for logging)\n job_id: Job ID for logging\n start: Start time for elapsed time calculation\n \"\"\"\n util.logger.info(f\"job {job_id}: handling image {path}\")\n \n # Create or get thumbnail instance\n thumbnail, _ = Thumbnail.objects.get_or_create(photo=photo)\n thumbnail._generate_thumbnail()\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: generate thumbnails: {path}, elapsed: {elapsed}\"\n )\n \n thumbnail._calculate_aspect_ratio()\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: calculate aspect ratio: {path}, elapsed: {elapsed}\"\n )\n \n # Calculate perceptual hash for duplicate detection\n if thumbnail.thumbnail_big and os.path.exists(thumbnail.thumbnail_big.path):\n phash = calculate_hash_from_thumbnail(thumbnail.thumbnail_big.path)\n if phash:\n photo.perceptual_hash = phash\n photo.save(update_fields=[\"perceptual_hash\"])\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: calculate perceptual hash: {path}, elapsed: {elapsed}\"\n )\n \n from api.models.photo_metadata import PhotoMetadata\n PhotoMetadata.extract_exif_data(photo, commit=True)\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: extract exif data: {path}, elapsed: {elapsed}\"\n )\n\n photo._extract_date_time_from_exif(True)\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: extract date time: {path}, elapsed: {elapsed}\"\n )\n \n thumbnail._get_dominant_color()\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: get dominant color: {path}, elapsed: {elapsed}\"\n )\n \n search_instance, created = PhotoSearch.objects.get_or_create(photo=photo)\n search_instance.recreate_search_captions()\n search_instance.save()\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n f\"job {job_id}: search caption recreated: {path}, elapsed: {elapsed}\"\n )\n" }, { "path": "api/directory_watcher/processing_jobs.py", "content": "\"\"\"\nPhoto processing jobs (tags, geolocation, faces).\n\nThese jobs run after the main scan to enrich photos with additional\nmetadata like location information, image tags, and face detection.\n\"\"\"\n\nimport traceback\nimport uuid\nfrom uuid import UUID\n\nfrom django import db\nfrom django.db.models import Q\nfrom django_q.tasks import AsyncTask\n\nfrom api import util\nfrom api.face_classify import cluster_all_faces\nfrom api.models import Face, LongRunningJob, Photo\nfrom api.models.photo_caption import PhotoCaption\nfrom api.directory_watcher.utils import update_scan_counter\n\n\ndef generate_face_embeddings(user, job_id: UUID):\n \"\"\"\n Generate face embeddings for faces that don't have them yet.\n \n Args:\n user: The user whose faces to process\n job_id: Job ID for tracking progress\n \"\"\"\n if Face.objects.filter(encoding=\"\").count() == 0:\n return\n \n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_GENERATE_FACE_EMBEDDINGS,\n job_id=job_id,\n )\n\n try:\n faces = Face.objects.filter(encoding=\"\")\n lrj.update_progress(current=0, target=faces.count())\n db.connections.close_all()\n\n for face in faces:\n failed = False\n error = None\n try:\n face.generate_encoding()\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n print(f\"[ERR]: {err}\")\n failed = True\n error_msg = f\"Face {face.id}: {str(err)}\\n{traceback.format_exc()}\"\n error = error_msg\n update_scan_counter(job_id, failed, error)\n\n lrj.complete()\n\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n print(f\"[ERR]: {err}\")\n lrj.fail(error=err)\n\n\ndef generate_tags(user, job_id: UUID, full_scan=False):\n \"\"\"\n Generate image tags (Places365 captions) for photos.\n \n Args:\n user: The user whose photos to process\n job_id: Job ID for tracking progress\n full_scan: If True, process all photos; otherwise only new ones\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_GENERATE_TAGS,\n job_id=job_id,\n )\n\n try:\n last_scan = (\n LongRunningJob.objects.filter(finished=True)\n .filter(job_type=LongRunningJob.JOB_GENERATE_TAGS)\n .filter(started_by=user)\n .order_by(\"-finished_at\")\n .first()\n )\n from constance import config as site_config\n\n tagging_model = site_config.TAGGING_MODEL\n\n existing_photos = Photo.objects.filter(\n Q(owner=user.id)\n & (\n Q(caption_instance__isnull=True)\n | Q(caption_instance__captions_json__isnull=True)\n | Q(**{f\"caption_instance__captions_json__{tagging_model}__isnull\": True})\n )\n )\n if not full_scan and last_scan:\n existing_photos = existing_photos.filter(added_on__gt=last_scan.started_at)\n\n if existing_photos.count() == 0:\n lrj.update_progress(current=0, target=0)\n lrj.complete()\n return\n lrj.update_progress(current=0, target=existing_photos.count())\n db.connections.close_all()\n\n for photo in existing_photos:\n AsyncTask(generate_tag_job, photo, job_id).run()\n\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n print(f\"[ERR]: {err}\")\n lrj.fail(error=err)\n\n\ndef generate_tag_job(photo: Photo, job_id: str):\n \"\"\"\n Worker task to generate tags for a single photo.\n \n Args:\n photo: The photo to process\n job_id: Job ID for tracking progress\n \"\"\"\n failed = False\n error = None\n try:\n photo.refresh_from_db()\n caption_instance, created = PhotoCaption.objects.get_or_create(photo=photo)\n caption_instance.generate_tag_captions(commit=True)\n except Exception as err:\n util.logger.exception(\"An error occurred: %s\", photo.image_hash)\n print(f\"[ERR]: {err}\")\n failed = True\n error_msg = f\"Photo {photo.image_hash}: {str(err)}\\n{traceback.format_exc()}\"\n error = error_msg\n update_scan_counter(job_id, failed, error)\n\n\ndef add_geolocation(user, job_id: UUID, full_scan=False):\n \"\"\"\n Add geolocation data to photos based on GPS coordinates.\n \n Args:\n user: The user whose photos to process\n job_id: Job ID for tracking progress\n full_scan: If True, process all photos; otherwise only new ones\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_ADD_GEOLOCATION,\n job_id=job_id,\n )\n\n try:\n last_scan = (\n LongRunningJob.objects.filter(finished=True)\n .filter(job_type=LongRunningJob.JOB_ADD_GEOLOCATION)\n .filter(started_by=user)\n .order_by(\"-finished_at\")\n .first()\n )\n existing_photos = Photo.objects.filter(owner=user.id)\n if not full_scan and last_scan:\n existing_photos = existing_photos.filter(added_on__gt=last_scan.started_at)\n if existing_photos.count() == 0:\n lrj.update_progress(current=0, target=0)\n lrj.complete()\n return\n lrj.update_progress(current=0, target=existing_photos.count())\n db.connections.close_all()\n\n for photo in existing_photos:\n AsyncTask(geolocation_job, photo, job_id).run()\n\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n print(f\"[ERR]: {err}\")\n lrj.fail(error=err)\n\n\ndef geolocation_job(photo: Photo, job_id: UUID):\n \"\"\"\n Worker task to add geolocation for a single photo.\n \n Args:\n photo: The photo to process\n job_id: Job ID for tracking progress\n \"\"\"\n failed = False\n error = None\n try:\n photo.refresh_from_db()\n photo._geolocate()\n photo._add_location_to_album_dates()\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n failed = True\n error_msg = f\"Photo {photo.image_hash}: {str(err)}\\n{traceback.format_exc()}\"\n error = error_msg\n update_scan_counter(job_id, failed, error)\n\n\ndef scan_faces(user, job_id: UUID, full_scan=False):\n \"\"\"\n Detect and extract faces from photos.\n \n Args:\n user: The user whose photos to process\n job_id: Job ID for tracking progress\n full_scan: If True, process all photos; otherwise only new ones\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_SCAN_FACES,\n job_id=job_id,\n )\n\n try:\n last_scan = (\n LongRunningJob.objects.filter(finished=True)\n .filter(job_type=LongRunningJob.JOB_SCAN_FACES)\n .filter(started_by=user)\n .order_by(\"-finished_at\")\n .first()\n )\n existing_photos = Photo.objects.filter(\n Q(owner=user.id) & Q(thumbnail__thumbnail_big__isnull=False)\n )\n if not full_scan and last_scan:\n existing_photos = existing_photos.filter(added_on__gt=last_scan.started_at)\n\n if existing_photos.count() == 0:\n lrj.update_progress(current=0, target=0)\n lrj.complete()\n return\n\n lrj.update_progress(current=0, target=existing_photos.count())\n db.connections.close_all()\n\n for photo in existing_photos:\n failed = False\n error = None\n try:\n photo._extract_faces()\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n print(f\"[ERR]: {err}\")\n failed = True\n error_msg = f\"Photo {photo.image_hash}: {str(err)}\\n{traceback.format_exc()}\"\n error = error_msg\n update_scan_counter(job_id, failed, error)\n except Exception as err:\n util.logger.exception(\"An error occurred: \")\n print(f\"[ERR]: {err}\")\n lrj.fail(error=err)\n\n generate_face_embeddings(user, uuid.uuid4())\n cluster_all_faces(user, uuid.uuid4())\n" }, { "path": "api/directory_watcher/repair_jobs.py", "content": "\"\"\"\nRepair jobs for fixing ungrouped file variants.\n\nThis module contains jobs that repair data inconsistencies, such as\nRAW files that weren't properly grouped with their JPEG counterparts\ndue to race conditions in previous scans.\n\"\"\"\n\nfrom uuid import UUID\n\nfrom api import util\nfrom api.models import File, LongRunningJob, Photo\nfrom api.directory_watcher.file_grouping import find_matching_jpeg_photo\n\n\ndef repair_ungrouped_file_variants(user, job_id: UUID):\n \"\"\"\n Post-scan job to fix any ungrouped file variants.\n \n This handles:\n 1. Race conditions from previous scans where RAW+JPEG weren't grouped\n 2. Rescans where files were added incrementally\n 3. Any orphaned RAW/metadata files that should be attached to existing Photos\n \n Strategy: Find Photos with RAW-only main_file, look for matching JPEG Photos,\n merge the RAW file into the JPEG Photo and delete the RAW-only Photo.\n \n Args:\n user: The user whose photos to repair\n job_id: Job ID for tracking progress\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_REPAIR_FILE_VARIANTS,\n job_id=job_id,\n )\n \n try:\n # Find Photos where main_file is RAW (potential orphans)\n raw_only_photos = Photo.objects.filter(\n owner=user,\n main_file__type=File.RAW_FILE\n )\n \n lrj.update_progress(current=0, target=raw_only_photos.count())\n \n merged_count = 0\n fixed_main_file_count = 0\n \n for raw_photo in raw_only_photos:\n if not raw_photo.main_file:\n continue\n \n # Check if this RAW photo has any IMAGE files (already grouped)\n has_image = raw_photo.files.filter(type=File.IMAGE).exists()\n if has_image:\n # Already properly grouped, just fix main_file priority\n image_file = raw_photo.files.filter(type=File.IMAGE).first()\n if image_file:\n raw_photo.main_file = image_file\n raw_photo.video = False\n raw_photo.save(update_fields=['main_file', 'video'])\n fixed_main_file_count += 1\n continue\n \n # Look for matching JPEG Photo\n jpeg_photo = find_matching_jpeg_photo(raw_photo.main_file.path, user)\n if jpeg_photo and jpeg_photo.id != raw_photo.id:\n # Merge: move all files from RAW photo to JPEG photo\n for f in raw_photo.files.all():\n if not jpeg_photo.files.filter(hash=f.hash).exists():\n jpeg_photo.files.add(f)\n \n jpeg_photo.save()\n \n # Delete the orphaned RAW-only photo\n raw_photo.delete()\n merged_count += 1\n util.logger.info(\n f\"job {job_id}: Merged RAW photo into JPEG photo {jpeg_photo.image_hash}\"\n )\n \n util.logger.info(\n f\"job {job_id}: Repaired {merged_count} ungrouped file variants, \"\n f\"fixed {fixed_main_file_count} main_file priorities\"\n )\n lrj.complete()\n \n except Exception as e:\n util.logger.exception(f\"job {job_id}: Error repairing file variants: {e}\")\n lrj.fail(error=e)\n" }, { "path": "api/directory_watcher/scan_jobs.py", "content": "\"\"\"\nMain scan jobs for photo discovery and processing.\n\nThis module contains the core scan_photos function that implements the\ntwo-phase scan architecture to avoid race conditions with RAW+JPEG grouping.\n\"\"\"\n\nimport datetime\nimport os\nimport uuid\nfrom collections import defaultdict\nfrom uuid import UUID\n\nimport pytz\nfrom django import db\nfrom django.conf import settings\nfrom django.core.paginator import Paginator\nfrom django.db.models import F, Q\nfrom django.utils import timezone\nfrom django_q.tasks import AsyncTask, Chain\n\nfrom api import util\nfrom api.metadata.reader import get_sidecar_files_in_priority_order\nfrom api.batch_jobs import batch_calculate_clip_embedding\nfrom api.models import LongRunningJob, Photo, Thumbnail\nfrom api.models.file import is_metadata\n\nfrom api.directory_watcher.file_grouping import get_file_grouping_key\nfrom api.directory_watcher.file_handlers import handle_new_image, handle_file_group\nfrom api.directory_watcher.processing_jobs import (\n generate_tags,\n add_geolocation,\n scan_faces,\n)\nfrom api.directory_watcher.repair_jobs import repair_ungrouped_file_variants\nfrom api.directory_watcher.utils import (\n walk_directory,\n walk_files,\n update_scan_counter,\n)\n\n\ndef _file_was_modified_after(filepath, time):\n \"\"\"Check if a file was modified after a given time.\"\"\"\n try:\n modified = os.path.getmtime(filepath)\n except OSError:\n return False\n return datetime.datetime.fromtimestamp(modified).replace(tzinfo=pytz.utc) > time\n\n\ndef wait_for_group_and_process_metadata(\n group_id: str,\n metadata_paths: list[str],\n user_id: int,\n full_scan: bool,\n job_id: UUID | str,\n expected_count: int,\n *,\n attempt: int = 1,\n max_attempts: int = 2,\n **kwargs, # Django-Q may pass additional arguments like 'schedule'\n):\n \"\"\"\n Sentinel task: waits until the expected number of image/video tasks in the group complete,\n then processes metadata files. It runs inside a django-q worker (non-blocking for the caller).\n\n Failure handling:\n - If the group is not complete yet, it will re-enqueue itself up to `max_attempts`.\n - After exhausting attempts, it proceeds with metadata processing anyway (best-effort).\n \"\"\"\n from django_q.tasks import count_group\n from django.contrib.auth import get_user_model\n\n util.logger.info(\n f\"Sentinel attempt {attempt}/{max_attempts} for group {group_id} (expecting {expected_count} tasks)\"\n )\n\n # Check current completion count for the group\n try:\n completed = count_group(group_id) # counts successes by default\n except Exception as e:\n util.logger.warning(\n f\"Could not read group status for {group_id}: {e}. Treating as incomplete.\"\n )\n completed = 0\n\n # Normalize to an int to avoid None-related type issues\n completed_int = int(completed or 0)\n\n if completed_int < expected_count and attempt < max_attempts:\n util.logger.info(\n f\"Group {group_id} not complete yet: {completed_int}/{expected_count}. Re-enqueue sentinel (attempt {attempt + 1}).\"\n )\n # Requeue the sentinel to check again later\n AsyncTask(\n wait_for_group_and_process_metadata,\n group_id,\n metadata_paths,\n user_id,\n full_scan,\n job_id,\n expected_count,\n attempt=attempt + 1,\n max_attempts=max_attempts,\n schedule=datetime.timedelta(seconds=5),\n ).run()\n return\n\n # Proceed with metadata processing (either completed or after exhausting attempts)\n if completed_int < expected_count:\n util.logger.warning(\n f\"Proceeding with metadata despite incomplete image group {group_id}: {completed_int}/{expected_count}.\"\n )\n else:\n util.logger.info(\n f\"Image group {group_id} completed. Processing {len(metadata_paths)} metadata files\"\n )\n\n if not metadata_paths:\n util.logger.info(\"No metadata files to process after images completion\")\n return\n\n User = get_user_model()\n try:\n user = User.objects.get(id=user_id)\n except User.DoesNotExist:\n util.logger.warning(\n f\"User {user_id} not found when processing metadata for job {job_id}\"\n )\n return\n\n last_scan = (\n LongRunningJob.objects.filter(finished=True)\n .filter(job_type=LongRunningJob.JOB_SCAN_PHOTOS)\n .filter(started_by=user)\n .order_by(\"-finished_at\")\n .first()\n )\n\n for path in metadata_paths:\n try:\n photo_scanner(user, last_scan, full_scan, path, job_id)\n except Exception as e:\n util.logger.exception(\n f\"Failed processing metadata {path} for job {job_id}: {e}\"\n )\n\n\ndef photo_scanner(user, last_scan, full_scan, path, job_id):\n \"\"\"\n Check if a single file needs processing and queue it.\n\n Used primarily for metadata files after the main scan.\n \"\"\"\n files_to_check = [path]\n files_to_check.extend(get_sidecar_files_in_priority_order(path))\n if (\n not Photo.objects.filter(files__path=path).exists()\n or full_scan\n or not last_scan\n or any(\n [_file_was_modified_after(p, last_scan.finished_at) for p in files_to_check]\n )\n ):\n # Queue processing for this file. Metadata is queued here without grouping on purpose,\n # because grouping is managed at the higher-level scan phase to ensure images complete first.\n AsyncTask(handle_new_image, user, path, job_id).run()\n else:\n update_scan_counter(job_id)\n\n\ndef scan_photos(user, full_scan, job_id, scan_directory=\"\", scan_files=[]):\n \"\"\"\n Two-phase scan to avoid race conditions with RAW+JPEG grouping.\n\n Phase 1: Collect all files and group by (directory, basename)\n - IMG_001.jpg, IMG_001.CR2, IMG_001.xmp -> one group\n - IMG_002.jpg -> separate group\n\n Phase 2: Process each group sequentially, creating one Photo per group\n with all file variants attached.\n\n This eliminates the race condition where concurrent processing of\n RAW and JPEG files could create separate Photos.\n\n Args:\n user: The user performing the scan\n full_scan: If True, rescan all files; otherwise only new/modified\n job_id: Job ID for tracking progress\n scan_directory: Directory to scan (defaults to user's scan_directory)\n scan_files: Optional list of specific files to scan\n \"\"\"\n thumbnail_dirs = [\n os.path.join(settings.MEDIA_ROOT, \"square_thumbnails_small\"),\n os.path.join(settings.MEDIA_ROOT, \"square_thumbnails\"),\n os.path.join(settings.MEDIA_ROOT, \"thumbnails_big\"),\n ]\n for directory in thumbnail_dirs:\n os.makedirs(directory, exist_ok=True)\n\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_SCAN_PHOTOS,\n job_id=job_id,\n )\n photo_count_before = Photo.objects.count()\n\n try:\n if scan_directory == \"\":\n scan_directory = user.scan_directory\n photo_list = []\n if scan_files:\n walk_files(scan_files, photo_list)\n else:\n walk_directory(scan_directory, photo_list)\n files_found = len(photo_list)\n last_scan = (\n LongRunningJob.objects.filter(finished=True)\n .filter(job_type=1)\n .filter(started_by=user)\n .order_by(\"-finished_at\")\n .first()\n )\n\n # === PHASE 1: Group files by (directory, basename) ===\n # This ensures RAW+JPEG pairs are processed together, eliminating race conditions\n file_groups: dict[tuple[str, str], list[str]] = defaultdict(list)\n metadata_paths: list[str] = []\n\n for path in photo_list:\n if is_metadata(path):\n # Metadata files are processed after their parent photos exist\n metadata_paths.append(path)\n else:\n # Group by (directory, basename_lowercase)\n group_key = get_file_grouping_key(path)\n file_groups[group_key].append(path)\n\n # Determine which groups need processing\n groups_to_process: list[tuple[tuple[str, str], list[str]]] = []\n\n for group_key, paths in file_groups.items():\n # Check if any file in this group needs processing\n needs_processing = False\n\n for path in paths:\n files_to_check = [path]\n files_to_check.extend(get_sidecar_files_in_priority_order(path))\n\n if (\n not Photo.objects.filter(files__path=path).exists()\n or full_scan\n or not last_scan\n or any(\n [\n _file_was_modified_after(p, last_scan.finished_at)\n for p in files_to_check\n ]\n )\n ):\n needs_processing = True\n break\n\n if needs_processing:\n groups_to_process.append((group_key, paths))\n\n # Progress target is number of groups (not individual files)\n # Each group = one Photo with potentially multiple file variants\n total_groups = len(groups_to_process) + len(metadata_paths)\n lrj.update_progress(current=0, target=total_groups)\n db.connections.close_all()\n\n util.logger.info(\n f\"Grouped {files_found} files into {len(file_groups)} groups, {len(groups_to_process)} need processing\"\n )\n\n # === PHASE 2: Process each file group ===\n # Process groups sequentially to avoid race conditions\n # Each group creates one Photo with all file variants\n image_group_id = str(uuid.uuid4())\n\n for group_key, paths in groups_to_process:\n AsyncTask(\n handle_file_group,\n user,\n paths,\n job_id,\n group=image_group_id,\n ).run()\n\n # If there are only metadata files (no image groups queued), process metadata now\n if not groups_to_process and metadata_paths:\n util.logger.info(\n f\"No images to process, processing {len(metadata_paths)} metadata files directly\"\n )\n for path in metadata_paths:\n photo_scanner(user, last_scan, full_scan, path, job_id)\n\n # If there are images and metadata, enqueue a sentinel task that waits for the image group\n if groups_to_process and metadata_paths:\n util.logger.info(\n f\"Scheduling sentinel to process {len(metadata_paths)} metadata files after {len(groups_to_process)} image groups\"\n )\n AsyncTask(\n wait_for_group_and_process_metadata,\n image_group_id,\n metadata_paths,\n user.id,\n full_scan,\n job_id,\n len(groups_to_process),\n attempt=1,\n max_attempts=2,\n ).run()\n\n util.logger.info(f\"Scanned {files_found} files in : {scan_directory}\")\n\n # If no files were queued for processing (empty directory or all files already processed),\n # mark the job as finished immediately since progress_current will equal progress_target (both 0)\n LongRunningJob.objects.filter(\n job_id=job_id, progress_current=F(\"progress_target\")\n ).update(finished=True, finished_at=timezone.now())\n\n util.logger.info(\"Finished updating album things\")\n\n # Check for photos with missing aspect ratios but existing thumbnails\n photos_with_missing_aspect_ratio = Photo.objects.filter(\n Q(owner=user.id)\n & Q(thumbnail__isnull=False)\n & Q(thumbnail__thumbnail_big__isnull=False)\n & Q(thumbnail__aspect_ratio__isnull=True)\n )\n if photos_with_missing_aspect_ratio.exists():\n util.logger.info(\n f\"Found {photos_with_missing_aspect_ratio.count()} photos with missing aspect ratios\"\n )\n for photo in photos_with_missing_aspect_ratio:\n try:\n thumbnail = getattr(photo, \"thumbnail\", None)\n if thumbnail and isinstance(thumbnail, Thumbnail):\n thumbnail._calculate_aspect_ratio()\n except Exception as e:\n util.logger.exception(\n f\"Could not calculate aspect ratio for photo {photo.image_hash}: {str(e)}\"\n )\n\n # if the scan type is not the default user scan directory, or if it is specified as only scanning\n # specific files, there is no need to rescan fully for missing photos.\n if full_scan or (scan_directory == user.scan_directory and not scan_files):\n AsyncTask(scan_missing_photos, user, uuid.uuid4()).run()\n\n # Run repair job to fix any previously ungrouped file variants\n # This handles race conditions from previous scans and incremental adds\n AsyncTask(repair_ungrouped_file_variants, user, uuid.uuid4()).run()\n\n AsyncTask(generate_tags, user, uuid.uuid4(), full_scan).run()\n AsyncTask(add_geolocation, user, uuid.uuid4(), full_scan).run()\n\n # The scan faces job will have issues if the embeddings haven't been generated before it runs\n chain = Chain()\n chain.append(batch_calculate_clip_embedding, user)\n chain.append(scan_faces, user, uuid.uuid4(), full_scan)\n chain.run()\n\n except Exception as e:\n util.logger.exception(\"An error occurred: \")\n lrj.fail(error=e)\n\n added_photo_count = Photo.objects.count() - photo_count_before\n util.logger.info(f\"Added {added_photo_count} photos\")\n\n\ndef scan_missing_photos(user, job_id: UUID):\n \"\"\"\n Scan for photos whose files no longer exist on disk.\n\n Args:\n user: The user whose photos to check\n job_id: Job ID for tracking progress\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_SCAN_MISSING_PHOTOS,\n job_id=job_id,\n )\n try:\n existing_photos = Photo.objects.filter(owner=user.id).order_by(\"image_hash\")\n\n paginator = Paginator(existing_photos, 5000)\n lrj.update_progress(current=0, target=paginator.num_pages)\n for page in range(1, paginator.num_pages + 1):\n for existing_photo in paginator.page(page).object_list:\n existing_photo._check_files()\n\n update_scan_counter(job_id)\n\n util.logger.info(\"Finished checking paths for missing photos\")\n except Exception as e:\n util.logger.exception(\"An error occurred: \")\n lrj.fail(error=e)\n" }, { "path": "api/directory_watcher/utils.py", "content": "\"\"\"\nUtility functions for directory scanning and job management.\n\"\"\"\n\nimport os\nimport stat\n\nfrom constance import config as site_config\nfrom django.db.models import F\nfrom django.utils import timezone\n\nfrom api.models import LongRunningJob\n\n\ndef should_skip(path):\n \"\"\"Check if a path should be skipped based on configured patterns.\"\"\"\n if not site_config.SKIP_PATTERNS:\n return False\n\n skip_patterns = site_config.SKIP_PATTERNS\n skip_list = skip_patterns.split(\",\")\n skip_list = map(str.strip, skip_list)\n\n res = [ele for ele in skip_list if (ele in path)]\n return bool(res)\n\n\nif os.name == \"Windows\":\n\n def is_hidden(path):\n \"\"\"Check if a file is hidden (Windows version).\"\"\"\n name = os.path.basename(os.path.abspath(path))\n return name.startswith(\".\") or _has_hidden_attribute(path)\n\n def _has_hidden_attribute(path):\n \"\"\"Check if file has Windows hidden attribute.\"\"\"\n try:\n return bool(os.stat(path).st_file_attributes & stat.FILE_ATTRIBUTE_HIDDEN)\n except Exception:\n return False\n\nelse:\n\n def is_hidden(path):\n \"\"\"Check if a file is hidden (Unix version - starts with dot).\"\"\"\n return os.path.basename(path).startswith(\".\")\n\n\ndef walk_directory(directory, callback):\n \"\"\"\n Recursively walk a directory and collect file paths.\n \n Args:\n directory: Directory to scan\n callback: List to append file paths to\n \"\"\"\n for file in os.scandir(directory):\n fpath = os.path.join(directory, file)\n if not is_hidden(fpath) and not should_skip(fpath):\n if os.path.isdir(fpath):\n walk_directory(fpath, callback)\n else:\n callback.append(fpath)\n\n\ndef walk_files(scan_files, callback):\n \"\"\"\n Walk a list of specific files.\n \n Args:\n scan_files: List of file paths to check\n callback: List to append valid file paths to\n \"\"\"\n for fpath in scan_files:\n if os.path.isfile(fpath):\n callback.append(fpath)\n\n\ndef update_scan_counter(job_id, failed=False, error=None):\n \"\"\"\n Update the progress counter for a long-running job.\n \n Increments progress_current and marks job as finished when complete.\n Also tracks errors for failed items.\n \n Args:\n job_id: The job ID to update\n failed: Whether this item failed processing\n error: Error message if failed\n \"\"\"\n # Increment the current progress and get the updated job\n LongRunningJob.objects.filter(job_id=job_id).update(\n progress_current=F(\"progress_current\") + 1\n )\n \n # Refetch the job to get the updated progress_current value\n job = LongRunningJob.objects.filter(job_id=job_id).first()\n if not job:\n return\n\n # Mark the job as finished if the current progress equals the target\n if job.progress_current >= job.progress_target:\n # Job is finishing, update result with errors if any\n result = job.result or {}\n if failed or error:\n result[\"status\"] = \"failed\"\n if \"errors\" not in result:\n result[\"errors\"] = []\n if error:\n error_str = str(error)\n # Avoid duplicate errors\n if error_str not in result[\"errors\"]:\n result[\"errors\"].append(error_str)\n # Set main error field for backward compatibility\n if \"error\" not in result and error:\n result[\"error\"] = str(error)\n elif \"error\" not in result and result.get(\"errors\"):\n result[\"error\"] = result[\"errors\"][0] # Use first error as main error\n job.finished = True\n job.finished_at = timezone.now()\n if failed:\n job.failed = True\n job.result = result\n job.save(update_fields=[\"finished\", \"finished_at\", \"failed\", \"result\"])\n else:\n # Job is still running, accumulate errors in result\n if failed or error:\n job = LongRunningJob.objects.filter(job_id=job_id).first()\n if job:\n result = job.result or {}\n result[\"status\"] = \"partial_failure\" if not job.finished else \"failed\"\n if \"errors\" not in result:\n result[\"errors\"] = []\n if error:\n error_str = str(error)\n # Avoid duplicate errors (limit to last 100 to prevent unbounded growth)\n if error_str not in result[\"errors\"]:\n result[\"errors\"].append(error_str)\n if len(result[\"errors\"]) > 100:\n result[\"errors\"] = result[\"errors\"][-100:] # Keep last 100 errors\n # Set main error field for backward compatibility\n if \"error\" not in result and error:\n result[\"error\"] = str(error)\n elif \"error\" not in result and result.get(\"errors\"):\n result[\"error\"] = result[\"errors\"][0] # Use first error as main error\n job.result = result\n job.failed = failed or job.failed\n job.save(update_fields=[\"failed\", \"result\"])\n" }, { "path": "api/drf_optimize.py", "content": "from django.db import ProgrammingError, models\nfrom django.db.models.constants import LOOKUP_SEP\nfrom django.db.models.query import normalize_prefetch_lookups\nfrom rest_framework import serializers\nfrom rest_framework.utils import model_meta\n\n\nclass OptimizeRelatedModelViewSetMetaclass(type):\n \"\"\"This metaclass optimizes the queryset using `prefetch_related` and `select_related`.\n\n Any attribute of `_base_forward_rel` as attributes on either the class or on\n any of its superclasses will be include in the `base_forward_rel`\n they must be ForeignKey fields and only those will be included.\n\n If the `serializer_class` attribute is an instance of `serializers.ModelSerializer`\n included as an attribute on the class the `serializers.ModelSerializer.Meta.fields`\n is also added calling prefetch_related on Many-To-One and Many-To-Many related objects.\n \"\"\"\n\n @classmethod\n def get_many_to_many_rel(cls, info, meta_fields):\n many_to_many_fields = [\n field_name\n for field_name, relation_info in info.relations.items()\n if relation_info.to_many\n ]\n many_to_many_lookups = []\n for lookup_name, lookup in cls.get_lookups(meta_fields):\n if lookup_name in many_to_many_fields:\n many_to_many_lookups.append(lookup)\n\n return many_to_many_lookups\n\n @classmethod\n def get_lookups(cls, fields, strict=False):\n field_lookups = [(lookup.split(LOOKUP_SEP, 1)[0], lookup) for lookup in fields]\n if strict:\n field_lookups = [f for f in field_lookups if LOOKUP_SEP in f[1]]\n return field_lookups\n\n @classmethod\n def get_many_to_one_rel(cls, info, meta_fields):\n try:\n fields = [\n field_name\n for field_name, relation_info in info.forward_relations.items()\n if issubclass(type(relation_info[0]), models.ForeignKey)\n ]\n except IndexError:\n pass\n else:\n if fields:\n forward_many_to_many_rel = []\n for lookup_name, lookup in cls.get_lookups(meta_fields, strict=True):\n if lookup_name in fields:\n forward_many_to_many_rel.append(lookup)\n return forward_many_to_many_rel\n return []\n\n @classmethod\n def get_forward_rel(cls, info, meta_fields):\n return [\n field_name\n for field_name, relation_info in info.forward_relations.items()\n if field_name in meta_fields and not relation_info.to_many\n ]\n\n def __new__(cls, name, bases, attrs):\n serializer_class = attrs.get(\"serializer_class\", None)\n many_to_many_fields = many_to_one_fields = related_fields = []\n\n info = None\n base_forward_rel = list(attrs.pop(\"_base_forward_rel\", ()))\n\n for base in reversed(bases):\n if hasattr(base, \"_base_forward_rel\"):\n base_forward_rel.extend(list(base._base_forward_rel))\n if serializer_class and issubclass(\n serializer_class, serializers.ModelSerializer\n ):\n base_forward_rel.extend(\n list(getattr(serializer_class, \"_related_fields\", [])),\n )\n many_to_many_fields.extend(\n list(getattr(serializer_class, \"_many_to_many_fields\", [])),\n )\n many_to_one_fields.extend(\n list(getattr(serializer_class, \"_many_to_one_fields\", [])),\n )\n if hasattr(serializer_class.Meta, \"model\"):\n info = model_meta.get_field_info(serializer_class.Meta.model)\n meta_fields = list(serializer_class.Meta.fields)\n many_to_many_fields.extend(meta_fields)\n many_to_one_fields.extend(meta_fields)\n base_forward_rel.extend(meta_fields)\n\n if info is not None:\n many_to_many_fields = cls.get_many_to_many_rel(\n info, set(many_to_many_fields)\n )\n many_to_one_fields = cls.get_many_to_one_rel(info, set(many_to_one_fields))\n related_fields = cls.get_forward_rel(info, set(base_forward_rel))\n\n queryset = attrs.get(\"queryset\", None)\n try:\n if queryset:\n if many_to_many_fields:\n queryset = queryset.prefetch_related(\n *normalize_prefetch_lookups(\n set(many_to_many_fields + many_to_one_fields)\n ),\n )\n if related_fields:\n queryset = queryset.select_related(*related_fields)\n attrs[\"queryset\"] = queryset.all()\n except ProgrammingError:\n pass\n return super(OptimizeRelatedModelViewSetMetaclass, cls).__new__(\n cls, name, bases, attrs\n )\n" }, { "path": "api/duplicate_detection.py", "content": "\"\"\"\nDuplicate detection module for finding duplicate photos.\n\nHandles two types of duplicates:\n- EXACT_COPY: Files with identical MD5 hash (byte-for-byte copies)\n- VISUAL_DUPLICATE: Photos with similar perceptual hash\n\nThis is separate from stack detection (RAW+JPEG pairs, bursts, etc.)\nbecause duplicates are about storage cleanup, not photo organization.\n\nOptimized with BK-Tree for efficient visual duplicate detection.\n\nMemory Optimizations (v2):\n- detect_exact_copies: Uses database aggregation (GROUP BY) instead of loading\n all photos into memory. Only photo IDs are loaded, not full objects.\n- detect_visual_duplicates: Processes photos in configurable batches (default 10k).\n Uses two-pass algorithm: within-batch BK-Tree search, then cross-batch linear scan.\n Memory usage: O(batch_size) instead of O(total_photos).\n\nWith 300k photos:\n- Old: ~10GB+ RAM (all photos + files + large BK-Tree)\n- New: ~100-200MB RAM (batch + hash list only)\n\"\"\"\n\nfrom collections import defaultdict\n\nfrom django.db.models import Q\n\nfrom api.models import Photo\nfrom api.models.duplicate import Duplicate\nfrom api.models.file import File\nfrom api.models.long_running_job import LongRunningJob\nfrom api.perceptual_hash import DEFAULT_HAMMING_THRESHOLD, hamming_distance\nfrom api.util import logger\n\n\nclass BKTree:\n \"\"\"\n Burkhard-Keller Tree for efficient Hamming distance queries.\n\n Achieves O(log n) average case by pruning branches using triangle inequality.\n \"\"\"\n\n def __init__(self, distance_func):\n self.distance = distance_func\n self.root = None\n self.size = 0\n\n def add(self, item_id, item_hash):\n \"\"\"Add an item (id, hash) to the tree.\"\"\"\n self.size += 1\n if self.root is None:\n self.root = {\"id\": item_id, \"hash\": item_hash, \"children\": {}}\n return\n\n node = self.root\n while True:\n dist = self.distance(item_hash, node[\"hash\"])\n if dist in node[\"children\"]:\n node = node[\"children\"][dist]\n else:\n node[\"children\"][dist] = {\n \"id\": item_id,\n \"hash\": item_hash,\n \"children\": {},\n }\n break\n\n def search(self, query_hash, threshold):\n \"\"\"Find all items within threshold Hamming distance of query.\"\"\"\n if self.root is None:\n return []\n\n results = []\n candidates = [self.root]\n\n while candidates:\n node = candidates.pop()\n dist = self.distance(query_hash, node[\"hash\"])\n\n if dist <= threshold:\n results.append((node[\"id\"], dist))\n\n min_dist = max(0, dist - threshold)\n max_dist = dist + threshold\n\n for d, child in node[\"children\"].items():\n if min_dist <= d <= max_dist:\n candidates.append(child)\n\n return results\n\n\nclass UnionFind:\n \"\"\"Union-Find with path compression and union by rank.\"\"\"\n\n def __init__(self):\n self.parent = {}\n self.rank = {}\n\n def find(self, x):\n if x not in self.parent:\n self.parent[x] = x\n self.rank[x] = 0\n return x\n if self.parent[x] != x:\n self.parent[x] = self.find(self.parent[x])\n return self.parent[x]\n\n def union(self, x, y):\n px, py = self.find(x), self.find(y)\n if px == py:\n return\n if self.rank[px] < self.rank[py]:\n px, py = py, px\n self.parent[py] = px\n if self.rank[px] == self.rank[py]:\n self.rank[px] += 1\n\n def get_groups(self):\n groups = defaultdict(list)\n for item in self.parent:\n groups[self.find(item)].append(item)\n return [group for group in groups.values() if len(group) > 1]\n\n\ndef detect_exact_copies(user, progress_callback=None):\n \"\"\"\n Detect exact file copies for a user.\n\n Groups photos that have the same content hash. Uses database aggregation\n to efficiently find duplicate groups without loading all photos into memory.\n\n Memory optimized: Uses database GROUP BY instead of Python dictionaries.\n\n Args:\n user: The user whose photos to analyze\n progress_callback: Optional callback(current, total, found) for progress\n\n Returns:\n Number of duplicate groups created\n \"\"\"\n from django.db.models import Count\n\n # Method 1: Find duplicate groups by Photo.image_hash using database aggregation\n # This is memory efficient as we only load photo IDs grouped by hash\n image_hash_groups = (\n Photo.objects.filter(\n Q(owner=user)\n & Q(hidden=False)\n & Q(in_trashcan=False)\n & Q(removed=False)\n & Q(image_hash__isnull=False)\n )\n .values(\"image_hash\")\n .annotate(count=Count(\"id\"))\n .filter(count__gt=1)\n .values_list(\"image_hash\", flat=True)\n )\n\n # Method 2: Find duplicate groups by File content hash (MD5 part)\n # We need to use a SUBSTRING operation to extract the MD5 part\n # This is done via raw SQL for efficiency\n from django.db import connection\n\n file_hash_duplicates = []\n with connection.cursor() as cursor:\n # Extract first 32 chars (MD5) from File.hash and find duplicates\n # Only consider non-metadata files\n cursor.execute(\n \"\"\"\n SELECT SUBSTRING(f.hash, 1, 32) as content_hash\n FROM api_file f\n INNER JOIN api_photo_files pf ON pf.file_id = f.hash\n INNER JOIN api_photo p ON p.id = pf.photo_id\n WHERE p.owner_id = %s \n AND p.hidden = FALSE \n AND p.in_trashcan = FALSE\n AND p.removed = FALSE\n AND f.type != %s\n GROUP BY SUBSTRING(f.hash, 1, 32)\n HAVING COUNT(DISTINCT p.id) > 1\n \"\"\",\n [user.id, File.METADATA_FILE],\n )\n\n file_hash_duplicates = [row[0] for row in cursor.fetchall()]\n\n # Use Union-Find to merge overlapping groups\n uf = UnionFind()\n\n # Process image_hash groups\n # Note: We need to iterate through the queryset, which will load the hashes into memory\n # But this is much better than loading all photos with files\n image_hash_list = list(image_hash_groups) # Load just the hashes\n total_image_groups = len(image_hash_list)\n\n for i, image_hash in enumerate(image_hash_list):\n # Only load photo IDs, not full Photo objects\n photo_ids = list(\n Photo.objects.filter(\n owner=user,\n image_hash=image_hash,\n hidden=False,\n in_trashcan=False,\n removed=False,\n ).values_list(\"id\", flat=True)\n )\n\n if len(photo_ids) >= 2:\n first = photo_ids[0]\n for pid in photo_ids[1:]:\n uf.union(first, pid)\n\n if progress_callback and i % 100 == 0:\n progress_callback(i, total_image_groups * 2, 0)\n\n # Process file_hash groups\n for i, content_hash in enumerate(file_hash_duplicates):\n # Find photos with files matching this content hash\n photo_ids = list(\n Photo.objects.filter(\n owner=user,\n hidden=False,\n in_trashcan=False,\n removed=False,\n files__hash__startswith=content_hash,\n )\n .exclude(files__type=File.METADATA_FILE)\n .distinct()\n .values_list(\"id\", flat=True)\n )\n\n if len(photo_ids) >= 2:\n first = photo_ids[0]\n for pid in photo_ids[1:]:\n uf.union(first, pid)\n\n if progress_callback and i % 100 == 0:\n progress_callback(total_image_groups + i, total_image_groups * 2, 0)\n\n # Get merged groups from Union-Find\n merged_groups = uf.get_groups()\n\n duplicates_created = 0\n total = len(merged_groups)\n\n for i, photo_id_group in enumerate(merged_groups):\n if len(photo_id_group) < 2:\n continue\n\n # Get Photo objects for this group\n group_photos = Photo.objects.filter(id__in=photo_id_group)\n\n # Create or merge duplicate group using the helper method\n duplicate = Duplicate.create_or_merge(\n owner=user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=group_photos,\n )\n\n if duplicate:\n duplicates_created += 1\n\n if progress_callback and i % 100 == 0:\n progress_callback(i, total, duplicates_created)\n\n logger.info(\n f\"Exact copy detection for {user.username}: found {duplicates_created} duplicate groups\"\n )\n return duplicates_created\n\n\ndef detect_visual_duplicates(\n user, threshold=DEFAULT_HAMMING_THRESHOLD, progress_callback=None, batch_size=10000\n):\n \"\"\"\n Detect visually similar photos using perceptual hash.\n\n Memory optimized: Processes photos in batches to avoid loading all data into memory.\n Uses a two-pass approach for complete duplicate detection with bounded memory.\n\n Algorithm:\n 1. First pass: Build BKTree in batches, find within-batch duplicates\n 2. Second pass: Compare each batch against all previous batches using linear scan\n\n The linear scan in pass 2 is acceptable because:\n - We only store (id, hash) tuples, not full Photo objects\n - Hamming distance is very fast to compute\n - With 300k photos, we have ~300k comparisons per batch, which is fast\n\n Args:\n user: The user whose photos to analyze\n threshold: Hamming distance threshold (default: 10)\n progress_callback: Optional callback(current, total, found) for progress\n batch_size: Number of photos to process per batch (default: 10000)\n\n Returns:\n Number of duplicate groups created\n \"\"\"\n # Get photos with perceptual hash that aren't already in visual duplicate groups\n # Exclude removed photos to avoid including merged/deleted duplicates\n photos_queryset = (\n Photo.objects.filter(\n Q(owner=user)\n & Q(hidden=False)\n & Q(in_trashcan=False)\n & Q(removed=False)\n & Q(perceptual_hash__isnull=False)\n )\n .exclude(duplicates__duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE)\n .only(\"id\", \"perceptual_hash\")\n )\n\n total = photos_queryset.count()\n if total < 2:\n return 0\n\n logger.info(\n f\"Processing {total} photos in batches of {batch_size} (user: {user.username})\"\n )\n\n # Union-Find for grouping across all batches\n uf = UnionFind()\n pairs_found = 0\n processed = 0\n\n # Store all photo hashes as (id, hash) tuples for cross-batch comparison\n # Memory efficient: 300k photos × ~28 bytes = ~8.4MB theoretical\n # In practice, Python overhead means ~25-40MB for list + objects\n all_photo_hashes = []\n\n # Calculate number of batches\n num_batches = (total + batch_size - 1) // batch_size\n\n # Pass 1: Process each batch internally and build the complete hash list\n for batch_idx in range(num_batches):\n offset = batch_idx * batch_size\n\n # Get current batch using slicing (memory efficient)\n batch_photos = list(\n photos_queryset[offset : offset + batch_size].values(\n \"id\", \"perceptual_hash\"\n )\n )\n\n if not batch_photos:\n break\n\n logger.info(\n f\"Pass 1: Processing batch {batch_idx + 1}/{num_batches} ({len(batch_photos)} photos)\"\n )\n\n # Build temporary BK-Tree for current batch (for efficient within-batch search)\n batch_tree = BKTree(hamming_distance)\n batch_hashes = []\n\n for photo in batch_photos:\n photo_id = photo[\"id\"]\n phash = photo[\"perceptual_hash\"]\n\n if phash:\n batch_tree.add(photo_id, phash)\n batch_hashes.append((photo_id, phash))\n\n # Find duplicates within current batch using BK-Tree\n for photo_id, phash in batch_hashes:\n similar = batch_tree.search(phash, threshold)\n\n for similar_id, distance in similar:\n if similar_id != photo_id:\n uf.union(photo_id, similar_id)\n pairs_found += 1\n\n # Add batch to the complete list for cross-batch comparison\n all_photo_hashes.extend(batch_hashes)\n\n processed += len(batch_photos)\n\n if progress_callback:\n # Report progress for pass 1 (first 50% of total work)\n progress_callback(processed // 2, total, pairs_found)\n\n logger.info(\n f\"Pass 1 complete. Found {pairs_found} within-batch pairs. Starting cross-batch comparison.\"\n )\n\n # Pass 2: Compare each batch against all previous photos (linear scan)\n # This ensures we don't miss duplicates between distant batches\n processed = 0\n\n for batch_idx in range(num_batches):\n start_idx = batch_idx * batch_size\n end_idx = min(start_idx + batch_size, len(all_photo_hashes))\n\n if start_idx >= end_idx:\n break\n\n batch_hashes = all_photo_hashes[start_idx:end_idx]\n\n logger.info(\n f\"Pass 2: Comparing batch {batch_idx + 1}/{num_batches} against previous photos\"\n )\n\n # Compare current batch against all previous photos\n # Store the previous photos slice once to avoid repeated slicing\n previous_hashes = all_photo_hashes[:start_idx] if start_idx > 0 else []\n\n for photo_id, phash in batch_hashes:\n # Only compare against photos in previous batches (avoid duplicate comparisons)\n for prev_id, prev_hash in previous_hashes:\n distance = hamming_distance(phash, prev_hash)\n if distance <= threshold:\n uf.union(photo_id, prev_id)\n pairs_found += 1\n\n processed += len(batch_hashes)\n\n if progress_callback:\n # Report progress for pass 2 (second 50% of total work)\n progress_callback(total // 2 + processed // 2, total, pairs_found)\n\n logger.info(f\"Pass 2 complete. Total pairs found: {pairs_found}\")\n\n # Create duplicate groups from Union-Find groups\n groups = uf.get_groups()\n duplicates_created = 0\n\n for group in groups:\n if len(group) < 2:\n continue\n\n # Get Photo objects for this group\n group_photos = Photo.objects.filter(id__in=group)\n\n # Create or merge duplicate group\n duplicate = Duplicate.create_or_merge(\n owner=user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n photos=group_photos,\n )\n\n if duplicate:\n duplicates_created += 1\n\n logger.info(\n f\"Visual duplicate detection for {user.username}: found {duplicates_created} groups from {pairs_found} pairs\"\n )\n return duplicates_created\n\n\ndef batch_detect_duplicates(user, options=None):\n \"\"\"\n Run batch duplicate detection for a user.\n\n Args:\n user: The user whose photos to analyze\n options: Dict with detection options:\n - detect_exact_copies: bool (default: True)\n - detect_visual_duplicates: bool (default: True)\n - visual_threshold: int (default: 10)\n - clear_pending: bool (default: False)\n - batch_size: int (default: 10000) - photos per batch for visual detection\n \"\"\"\n if options is None:\n options = {}\n\n detect_exact = options.get(\"detect_exact_copies\", True)\n detect_visual = options.get(\"detect_visual_duplicates\", True)\n visual_threshold = options.get(\"visual_threshold\", DEFAULT_HAMMING_THRESHOLD)\n clear_pending = options.get(\"clear_pending\", False)\n batch_size = options.get(\"batch_size\", 10000)\n\n # Create long-running job for progress tracking\n job = LongRunningJob.create_job(\n user=user,\n job_type=LongRunningJob.JOB_DETECT_DUPLICATES,\n start_now=True,\n )\n\n try:\n # Clear pending duplicates if requested\n if clear_pending:\n cleared = Duplicate.objects.filter(\n owner=user, review_status=Duplicate.ReviewStatus.PENDING\n ).delete()[0]\n logger.info(f\"Cleared {cleared} pending duplicates for {user.username}\")\n\n total_found = 0\n\n # Detect exact copies\n if detect_exact:\n\n def progress_exact(current, total, found):\n job.set_result(\n {\n \"stage\": \"exact_copies\",\n \"current\": current,\n \"total\": total,\n \"found\": found,\n }\n )\n\n exact_count = detect_exact_copies(user, progress_exact)\n total_found += exact_count\n\n # Detect visual duplicates\n if detect_visual:\n\n def progress_visual(current, total, found):\n job.set_result(\n {\n \"stage\": \"visual_duplicates\",\n \"current\": current,\n \"total\": total,\n \"found\": found,\n }\n )\n\n visual_count = detect_visual_duplicates(\n user, visual_threshold, progress_visual, batch_size\n )\n total_found += visual_count\n\n job.complete(result={\"status\": \"completed\", \"duplicates_found\": total_found})\n\n logger.info(\n f\"Duplicate detection completed for {user.username}: {total_found} groups found\"\n )\n\n except Exception as e:\n logger.error(f\"Duplicate detection failed for {user.username}: {e}\")\n job.fail(error=e)\n raise\n" }, { "path": "api/face_classify.py", "content": "import datetime\nimport uuid\n\nimport numpy as np\nimport seaborn as sns\nfrom bulk_update.helper import bulk_update\nfrom django.core.paginator import Paginator\nfrom django.db.models import Q\nfrom django_q.tasks import AsyncTask\nfrom hdbscan import HDBSCAN\nfrom sklearn.decomposition import PCA\nfrom sklearn.neural_network import MLPClassifier\n\nfrom api.cluster_manager import ClusterManager\nfrom api.models import Face, LongRunningJob, Person\nfrom api.models.cluster import UNKNOWN_CLUSTER_ID, Cluster, get_unknown_cluster\nfrom api.models.user import User, get_deleted_user\nfrom api.util import logger\n\nFACE_CLASSIFY_COLUMNS = [\n \"person\",\n \"classification_person\",\n \"classification_probability\",\n \"cluster_person\",\n \"cluster_probability\",\n \"id\",\n \"cluster\",\n]\n\n\ndef cluster_faces(user, inferred=True):\n # Fetch distinct persons associated with the user's faces\n persons = [p.id for p in Person.objects.filter(faces__photo__owner=user).distinct()]\n\n # Create a color mapping for each person\n p2c = dict(zip(persons, sns.color_palette(n_colors=len(persons)).as_hex()))\n\n face_encoding = []\n faces_with_encoding = []\n # Fetch faces that belong to the user and are not deleted\n faces = Face.objects.filter(Q(photo__owner=user) & Q(deleted=False))\n paginator = Paginator(faces, 5000)\n\n for page in range(1, paginator.num_pages + 1):\n for face in paginator.page(page).object_list:\n if ((not face.person) or inferred) and face.encoding:\n face_encoding.append(face.get_encoding_array())\n faces_with_encoding.append(face)\n\n # Return empty result if no faces with encodings\n if len(face_encoding) == 0:\n return {\"status\": True, \"data\": []}\n\n # Perform PCA for dimensionality reduction\n pca = PCA(n_components=3)\n vis_all = pca.fit_transform(face_encoding)\n\n res = []\n for face, vis in zip(faces_with_encoding, vis_all):\n person_id = face.person.id if face.person else UNKNOWN_CLUSTER_ID\n person_name = face.person.name if face.person else \"unknown\"\n\n # Ensure UNKNOWN_CLUSTER_ID is in p2c\n if person_id not in p2c:\n # Assign a default color if not found\n color = \"#000000\" # Default to black or any fallback color\n else:\n color = p2c[person_id]\n\n res.append(\n {\n \"person_id\": person_id,\n \"person_name\": person_name,\n \"person_label_is_inferred\": not face.person,\n \"color\": color,\n \"face_url\": face.image.url,\n \"value\": {\"x\": vis[0], \"y\": vis[1], \"size\": vis[2]},\n }\n )\n return {\"status\": True, \"data\": res}\n\n\ndef cluster_all_faces(user, job_id) -> bool:\n \"\"\"Groups all faces into clusters for ease of labeling. It first deletes all\n existing clusters, then regenerates them all. It will split clusters that have\n more than one kind of labeled face.\n :param user: the current user running the training\n :param job_id: the background job ID\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_CLUSTER_ALL_FACES,\n job_id=job_id,\n )\n lrj.update_progress(current=0, target=1)\n\n try:\n delete_clustered_people(user)\n delete_clusters(user)\n delete_persons_without_faces()\n target_count: int = create_all_clusters(user, lrj)\n\n lrj.update_progress(current=target_count, target=target_count)\n lrj.complete()\n\n train_job_id = uuid.uuid4()\n AsyncTask(train_faces, user, train_job_id).run()\n return True\n\n except BaseException as err:\n logger.exception(\"An error occurred\")\n print(f\"[ERR] {err}\")\n lrj.fail(error=err)\n return False\n\n\ndef create_all_clusters(user: User, lrj: LongRunningJob = None) -> int:\n \"\"\"Generate Cluster records for each different clustering of people\n :param user: the current user\n :param lrj: LongRunningJob to update, if needed\n \"\"\"\n all_clusters: list[Cluster] = []\n face: Face\n logger.info(\"Creating clusters\")\n\n data = {\n \"all\": {\"encoding\": [], \"id\": [], \"person_id\": [], \"person_labeled\": []},\n }\n for face in Face.objects.filter(photo__owner=user).prefetch_related(\"person\"):\n data[\"all\"][\"encoding\"].append(face.get_encoding_array())\n data[\"all\"][\"id\"].append(face.id)\n\n target_count = len(data[\"all\"][\"id\"])\n if target_count == 0:\n return target_count\n\n min_cluster_size = 2\n # double cluster size for every 10x increase in target counts, if user has not set a valid min_cluster_size\n if (\n user.min_cluster_size == 0\n or user.min_cluster_size is None\n or user.min_cluster_size == 1\n ):\n if target_count > 1000:\n min_cluster_size = 4\n if target_count > 10000:\n min_cluster_size = 8\n if target_count > 100000:\n min_cluster_size = 16\n else:\n min_cluster_size = user.min_cluster_size\n\n min_samples = 1\n if user.min_samples > 0:\n min_samples = user.min_samples\n\n # creating HDBSCAN object for clustering the encodings with the metric \"euclidean\"\n clt = HDBSCAN(\n min_cluster_size=min_cluster_size,\n min_samples=min_samples,\n cluster_selection_epsilon=user.cluster_selection_epsilon,\n metric=\"euclidean\",\n )\n logger.info(\"Before finding clusters\")\n # clustering the encodings\n clt.fit(np.array(data[\"all\"][\"encoding\"]))\n logger.info(\"After finding clusters\")\n\n labelIDs = np.unique(clt.labels_)\n labelID: np.intp\n commit_time = datetime.datetime.now() + datetime.timedelta(seconds=5)\n count: int = 0\n maxLen: int = len(str(np.size(labelIDs)))\n sortedIndexes: dict[int, np.ndarray] = dict()\n clusterCount: int = 0\n clusterId: int\n\n for labelID in labelIDs:\n idxs = np.where(clt.labels_ == labelID)[0]\n sortedIndexes[labelID] = idxs\n\n logger.info(f\"Found {len(sortedIndexes)} clusters\")\n for labelID in sorted(\n sortedIndexes, key=lambda key: np.size(sortedIndexes[key]), reverse=True\n ):\n if labelID != UNKNOWN_CLUSTER_ID:\n clusterCount = clusterCount + 1\n clusterId = clusterCount\n else:\n clusterId = labelID\n face_array: list[Face] = []\n face_id_list: list[int] = []\n for i in sortedIndexes[labelID]:\n count = count + 1\n face_id = data[\"all\"][\"id\"][i]\n face_id_list.append(face_id)\n face_array = Face.objects.filter(\n Q(pk__in=face_id_list) & Q(encoding__isnull=False) & Q(deleted=False)\n )\n new_clusters: list[Cluster] = ClusterManager.try_add_cluster(\n user, clusterId, face_array, maxLen\n )\n\n if commit_time < datetime.datetime.now() and lrj is not None:\n lrj.progress_current = count\n lrj.progress_target = target_count\n lrj.save()\n commit_time = datetime.datetime.now() + datetime.timedelta(seconds=5)\n\n all_clusters.extend(new_clusters)\n\n print(f\"[INFO] Created {len(all_clusters)} clusters\")\n return target_count\n\n\ndef delete_persons_without_faces():\n \"\"\"Delete all existing Person records that have no associated Face records\"\"\"\n print(\"[INFO] Deleting all people without faces\")\n Person.objects.filter(faces=None, kind=Person.KIND_USER).delete()\n\n\ndef delete_clusters(user: User):\n \"\"\"Delete all existing Cluster records\"\"\"\n print(\"[INFO] Deleting all clusters\")\n Cluster.objects.filter(Q(owner=user)).delete()\n Cluster.objects.filter(Q(owner=None)).delete()\n Cluster.objects.filter(Q(owner=get_deleted_user())).delete()\n\n\ndef delete_clustered_people(user: User):\n \"\"\"Delete all existing Person records of type CLUSTER\"\"\"\n print(\"[INFO] Deleting all clustered people\")\n Person.objects.filter(kind=Person.KIND_CLUSTER, cluster_owner=user).delete()\n Person.objects.filter(kind=Person.KIND_UNKNOWN, cluster_owner=user).delete()\n Person.objects.filter(cluster_owner=None).delete()\n Person.objects.filter(cluster_owner=get_deleted_user()).delete()\n\n\n# Function to filter data based on a desired shape\ndef filter_data(encodings, ids):\n valid_encodings = []\n valid_ids = []\n expected_shape = (\n len(encodings[0]) if encodings else 0\n ) # Set expected shape from first entry\n\n for i, (encoding, id_) in enumerate(zip(encodings, ids)):\n if len(encoding) == expected_shape: # Check if shape is consistent\n valid_encodings.append(encoding)\n valid_ids.append(id_)\n else:\n logger.error(\n f\"Discarding entry {i}: ID={id_}, encoding shape={len(encoding)} (expected {expected_shape})\"\n )\n\n return np.array(valid_encodings), np.array(valid_ids)\n\n\ndef train_faces(user: User, job_id) -> bool:\n \"\"\"Given existing Cluster records for all faces, determines the probability\n that unknown faces belong to those Clusters. It takes any known, labeled faces\n and adds the centroids of \"unknown\" clusters, assuming that those clusters\n correspond to *some* face. It then trains a classifier on that data to use\n in calculating the probabilities for unknown faces.\n :param user: the current user running the training\n :param job_id: the background job ID\n \"\"\"\n lrj = LongRunningJob.get_or_create_job(\n user=user,\n job_type=LongRunningJob.JOB_TRAIN_FACES,\n job_id=job_id,\n )\n lrj.update_progress(current=1, target=2)\n try:\n # Use two array, so that the first one gets thrown out, if it is no longer used.\n data_known = {\"encoding\": [], \"id\": []}\n data_unknown = {\"encoding\": [], \"id\": []}\n # First, sort all faces into known and unknown ones\n face: Face\n for face in Face.objects.filter(\n Q(photo__owner=user) & Q(encoding__isnull=False) & Q(deleted=False)\n ).prefetch_related(\"person\"):\n if not face.person:\n data_unknown[\"encoding\"].append(face.get_encoding_array())\n data_unknown[\"id\"].append(face.id)\n else:\n data_known[\"encoding\"].append(face.get_encoding_array())\n data_known[\"id\"].append(face.person.id)\n\n if len(data_known[\"id\"]) == 0:\n classifier = None\n else:\n logger.info(\"Before fitting\")\n classifier = MLPClassifier(\n solver=\"adam\", alpha=1e-5, random_state=1, max_iter=1000\n ).fit(np.array(data_known[\"encoding\"]), np.array(data_known[\"id\"]))\n logger.info(\"After fitting\")\n\n # Next, pretend all unknown face clusters are known and add their mean encoding. This allows us\n # to predict the likelihood of other unknown faces belonging to those simulated clusters. For\n # the \"Unknown - Other\"-type cluster, we can still try to predict the probability that the face\n # can't be classified into another group, i.e. that it should be classified that way\n cluster: Cluster\n for cluster in Cluster.objects.filter(owner=user):\n if cluster.person and cluster.person.kind == Person.KIND_CLUSTER:\n print(cluster.person)\n data_known[\"encoding\"].append(cluster.get_mean_encoding_array())\n data_known[\"id\"].append(cluster.person.id)\n\n filtered_encodings, filtered_ids = filter_data(\n data_known[\"encoding\"], data_known[\"id\"]\n )\n\n # Fit the classifier based on the \"known\" faces, including the simulated clusters\n logger.info(\"Before cluster fitting\")\n cluster_classifier = MLPClassifier(\n solver=\"adam\", alpha=1e-5, random_state=1, max_iter=1000\n ).fit(filtered_encodings, filtered_ids)\n logger.info(\"After cluster fitting\")\n\n # Collect the probabilities for each unknown face. The probabilities returned\n # are arrays in the same order as the people IDs in the original training set\n target_count = len(data_unknown[\"id\"])\n if target_count == 0:\n logger.info(\"No clusters found\")\n lrj.update_progress(current=2, target=2)\n lrj.complete()\n return True\n logger.info(f\"Number of Cluster: {target_count}\")\n\n # Hacky way to split arrays into smaller arrays\n pages_encoding = [\n data_unknown[\"encoding\"][i : i + 100]\n for i in range(0, len(data_unknown[\"encoding\"]), 100)\n ]\n pages_id = [\n data_unknown[\"id\"][i : i + 100]\n for i in range(0, len(data_unknown[\"encoding\"]), 100)\n ]\n for idx, page in enumerate(pages_encoding):\n page_id = pages_id[idx]\n pages_of_faces = Face.objects.filter(id__in=page_id).all()\n # sort pages of faces by id by page_id\n pages_of_faces = sorted(pages_of_faces, key=lambda x: page_id.index(x.id))\n face_encodings_unknown_np = np.array(page)\n cluster_probs = cluster_classifier.predict_proba(face_encodings_unknown_np)\n if classifier:\n classification_probs = classifier.predict_proba(\n face_encodings_unknown_np\n )\n else:\n classification_probs = []\n while len(classification_probs) < len(cluster_probs):\n classification_probs.append(\n [0.0] * len(cluster_classifier.classes_)\n )\n\n commit_time = datetime.datetime.now() + datetime.timedelta(seconds=5)\n face_stack = []\n unknown_cluster: Cluster = get_unknown_cluster(user=user)\n\n for idx, (\n face,\n cluster_probabilities,\n classification_probabilties,\n ) in enumerate(zip(pages_of_faces, cluster_probs, classification_probs)):\n face.cluster_probability = 0.0 # Cluster probability\n face.classification_probability = 0.0 # Classification probability\n classification_person = None\n classification_probability = 0.0\n\n highest_classification_probability = max(classification_probabilties)\n highest_classification_person = 0\n\n # Find the person with the highest probability for classification\n if classifier:\n for i, target in enumerate(classifier.classes_):\n if (\n highest_classification_probability\n == classification_probabilties[i]\n ):\n highest_classification_person = target\n\n classification_person = highest_classification_person\n classification_probability = highest_classification_probability\n\n # Find the probability in the probability array corresponding to the person\n # that we currently believe the face is, even a simulated \"unknown\" person\n highest_probability = max(cluster_probabilities)\n highest_probability_person = 0\n for i, target in enumerate(cluster_classifier.classes_):\n if highest_probability == cluster_probabilities[i]:\n highest_probability_person = target\n\n if face.cluster != unknown_cluster:\n face.cluster_person = Person.objects.get(\n id=highest_probability_person\n )\n face.cluster_probability = highest_probability\n\n if classification_person:\n face.classification_person = Person.objects.get(\n id=classification_person\n )\n face.classification_probability = classification_probability\n\n face_stack.append(face)\n if commit_time < datetime.datetime.now():\n lrj.update_progress(current=idx + 1, target=target_count)\n commit_time = datetime.datetime.now() + datetime.timedelta(\n seconds=5\n )\n if len(face_stack) > 200:\n bulk_update(face_stack, update_fields=FACE_CLASSIFY_COLUMNS)\n face_stack = []\n\n bulk_update(face_stack, update_fields=FACE_CLASSIFY_COLUMNS)\n\n lrj.update_progress(current=target_count, target=target_count)\n lrj.complete()\n return True\n\n except BaseException as err:\n logger.exception(\"An error occurred\")\n print(f\"[ERR] {err}\")\n lrj.fail(error=err)\n return False\n" }, { "path": "api/face_extractor.py", "content": "import numpy as np\nimport PIL\n\nfrom api.face_recognition import get_face_locations\nfrom api.metadata.reader import get_metadata\nfrom api.metadata.tags import Tags\nfrom api.util import is_number, logger\n\n\nclass RuleTypes:\n EXIF = \"exif\"\n DLIB = \"dlib\"\n\n\ndef extract_from_exif(image_path, big_thumbnail_image_path):\n (region_info, orientation) = get_metadata(\n image_path,\n tags=[Tags.REGION_INFO, Tags.ORIENTATION],\n try_sidecar=True,\n struct=True,\n )\n if not region_info:\n return\n logger.debug(f\"Extracted region_info for {image_path}\")\n logger.debug(f\"region_info: {region_info}\")\n face_locations = []\n for region in region_info[\"RegionList\"]:\n if region.get(\"Type\") != \"Face\":\n continue\n person_name = region.get(\"Name\")\n\n area = region.get(\"Area\")\n applied_to_dimensions = region.get(\"AppliedToDimensions\")\n big_thumbnail_image = np.array(PIL.Image.open(big_thumbnail_image_path))\n if (area and area.get(\"Unit\") == \"normalized\") or (\n applied_to_dimensions and applied_to_dimensions.get(\"Unit\") == \"pixel\"\n ):\n image_width = big_thumbnail_image.shape[1]\n image_height = big_thumbnail_image.shape[0]\n if (\n not is_number(area.get(\"X\"))\n or not is_number(area.get(\"Y\"))\n or not is_number(area.get(\"W\"))\n or not is_number(area.get(\"H\"))\n ):\n logger.info(\n f\"Broken face area exif data! No numerical positional data. region_info: {region_info}\"\n )\n continue\n\n correct_w = float(area.get(\"W\"))\n correct_h = float(area.get(\"H\"))\n correct_x = float(area.get(\"X\"))\n correct_y = float(area.get(\"Y\"))\n if orientation == \"Rotate 90 CW\":\n temp_x = correct_x\n correct_x = 1 - correct_y\n correct_y = temp_x\n correct_w, correct_h = correct_h, correct_w\n elif orientation == \"Mirror horizontal\":\n correct_x = 1 - correct_x\n elif orientation == \"Rotate 180\":\n correct_x = 1 - correct_x\n correct_y = 1 - correct_y\n elif orientation == \"Mirror vertical\":\n correct_y = 1 - correct_y\n elif orientation == \"Mirror horizontal and rotate 270 CW\":\n temp_x = correct_x\n correct_x = 1 - correct_y\n correct_y = temp_x\n correct_w, correct_h = correct_h, correct_w\n elif orientation == \"Mirror horizontal and rotate 90 CW\":\n temp_x = correct_x\n correct_x = correct_y\n correct_y = 1 - temp_x\n correct_w, correct_h = correct_h, correct_w\n elif orientation == \"Rotate 270 CW\":\n temp_x = correct_x\n correct_x = correct_y\n correct_y = 1 - temp_x\n correct_w, correct_h = correct_h, correct_w\n\n # Calculate the half-width and half-height of the box\n half_width = (correct_w * image_width) / 2\n half_height = (correct_h * image_height) / 2\n\n # Calculate the top, right, bottom, and left coordinates\n top = int((correct_y * image_height) - half_height)\n right = int((correct_x * image_width) + half_width)\n bottom = int((correct_y * image_height) + half_height)\n left = int((correct_x * image_width) - half_width)\n\n face_locations.append((top, right, bottom, left, person_name))\n return face_locations\n\n\ndef extract_from_dlib(image_path, big_thumbnail_path, owner):\n try:\n face_locations = get_face_locations(\n big_thumbnail_path,\n model=owner.face_recognition_model.lower(),\n )\n except Exception as e:\n logger.info(f\"Can't extract face information on photo: {image_path}\")\n logger.info(e)\n face_locations = []\n\n for i, face_location in enumerate(face_locations):\n face_locations[i] = (*face_location, None)\n return face_locations\n\n\ndef extract(image_path, big_thumbnail_path, owner):\n exif = extract_from_exif(image_path, big_thumbnail_path)\n if not exif:\n return extract_from_dlib(image_path, big_thumbnail_path, owner)\n return exif\n" }, { "path": "api/face_recognition.py", "content": "import numpy as np\nimport requests\n\n\ndef get_face_encodings(image_path, known_face_locations):\n json = {\n \"source\": image_path,\n \"face_locations\": known_face_locations,\n }\n face_encoding = requests.post(\n \"http://localhost:8005/face-encodings\", json=json\n ).json()\n\n face_encodings_list = face_encoding[\"encodings\"]\n face_encodings = [np.array(enc) for enc in face_encodings_list]\n\n return face_encodings\n\n\ndef get_face_locations(image_path, model=\"hog\"):\n json = {\"source\": image_path, \"model\": model}\n face_locations = requests.post(\n \"http://localhost:8005/face-locations\", json=json\n ).json()\n return face_locations[\"face_locations\"]\n" }, { "path": "api/feature/__init__.py", "content": "" }, { "path": "api/feature/embedded_media.py", "content": "import os\nfrom mmap import ACCESS_READ, mmap\n\nimport magic\nfrom django.conf import settings\n\nJPEG_EOI_MARKER = b\"\\xff\\xd9\"\nGOOGLE_PIXEL_MOTION_PHOTO_MP4_SIGNATURES = [b\"ftypmp42\", b\"ftypisom\", b\"ftypiso2\"]\n\n# in reality Samsung motion photo marker will look something like this\n# ........Image_UTC_Data1458170015363SEFHe...........#...#.......SEFT..0.....MotionPhoto_Data\n# but we are interested only in the content of the video which is right after MotionPhoto_Data\nSAMSUNG_MOTION_PHOTO_MARKER = b\"MotionPhoto_Data\"\n\n\ndef _locate_embedded_video_google(data):\n signatures = GOOGLE_PIXEL_MOTION_PHOTO_MP4_SIGNATURES\n for signature in signatures:\n position = data.find(signature)\n if position != -1:\n return position - 4\n return -1\n\n\ndef _locate_embedded_video_samsung(data):\n position = data.find(SAMSUNG_MOTION_PHOTO_MARKER)\n if position != -1:\n return position + len(SAMSUNG_MOTION_PHOTO_MARKER)\n return -1\n\n\ndef has_embedded_media(path: str) -> bool:\n mime = magic.Magic(mime=True)\n mime_type = mime.from_file(path)\n if mime_type != \"image/jpeg\":\n return False\n with open(path, \"rb\") as image:\n with mmap(image.fileno(), 0, access=ACCESS_READ) as mm:\n return (\n _locate_embedded_video_samsung(mm) != -1\n or _locate_embedded_video_google(mm) != -1\n )\n\n\ndef extract_embedded_media(path: str, hash: str) -> str | None:\n with open(str(path), \"rb\") as image:\n with mmap(image.fileno(), 0, access=ACCESS_READ) as mm:\n position = _locate_embedded_video_google(\n mm\n ) or _locate_embedded_video_google(mm)\n if position == -1:\n return None\n output_dir = f\"{settings.MEDIA_ROOT}/embedded_media\"\n if not os.path.exists(output_dir):\n os.makedirs(output_dir)\n output_path = f\"{output_dir}/{hash}_1.mp4\"\n with open(output_path, \"wb+\") as video:\n mm.seek(position)\n data = mm.read(mm.size())\n video.write(data)\n return output_path\n" }, { "path": "api/feature/tests/__init__.py", "content": "" }, { "path": "api/feature/tests/test_embedded_media.py", "content": "from django.conf import settings\nfrom django.test import override_settings\nfrom rest_framework.test import APIClient, APITestCase\n\nfrom api.feature.embedded_media import (\n GOOGLE_PIXEL_MOTION_PHOTO_MP4_SIGNATURES,\n JPEG_EOI_MARKER,\n SAMSUNG_MOTION_PHOTO_MARKER,\n extract_embedded_media,\n has_embedded_media,\n)\nfrom api.models import User\nfrom api.models.file import File\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\ndef create_test_file(path: str, user: User, content: bytes):\n with open(path, \"wb+\") as f:\n f.write(content)\n return File.create(path, user)\n\n\nJPEG_MAGIC_NUMBER = b\"\\xff\\xd8\\xff\"\nJPEG = JPEG_MAGIC_NUMBER + b\"\\xde\\xad\\xfa\\xce\" + JPEG_EOI_MARKER\nMP4_DATA = b\"\\xca\\xfe\\xfe\\xed\"\nMP4_PREFIX = b\"\\x00\\x00\\x00\\x18\"\nMP4 = MP4_PREFIX + b\"ftypmp42\" + MP4_DATA\nRANDOM_BYTES = b\"\\x13\\x37\\xc0\\xde\"\n\n\n@override_settings(MEDIA_ROOT=\"/tmp\")\nclass EmbeddedMediaTest(APITestCase):\n def setUp(self):\n self.test_image_path = \"/tmp/test_file.jpeg\"\n self.test_video_path = \"/tmp/test_file.mp4\"\n self.user = create_test_user()\n self.client = APIClient()\n\n def test_should_not_process_non_jpeg_files(self):\n file = create_test_file(self.test_video_path, self.user, MP4)\n actual = has_embedded_media(file.path)\n self.assertFalse(actual)\n\n def test_google_pixel_motion_photo_signatures(self):\n for signature in GOOGLE_PIXEL_MOTION_PHOTO_MP4_SIGNATURES:\n content = JPEG + MP4_PREFIX + signature + MP4_DATA\n file = create_test_file(self.test_image_path, self.user, content)\n actual = has_embedded_media(file.path)\n self.assertTrue(actual)\n\n def test_samsung_motion_photo_signature(self):\n content = JPEG + SAMSUNG_MOTION_PHOTO_MARKER + MP4_DATA\n file = create_test_file(self.test_image_path, self.user, content)\n actual = has_embedded_media(file.path)\n self.assertTrue(actual)\n\n def test_other_content_should_not_report_as_having_embedded_media(self):\n file = create_test_file(self.test_image_path, self.user, RANDOM_BYTES)\n actual = has_embedded_media(file.path)\n self.assertFalse(actual)\n\n def test_extract_embedded_media_from_google_motion_photo(self):\n for signature in GOOGLE_PIXEL_MOTION_PHOTO_MP4_SIGNATURES:\n content = JPEG + MP4_PREFIX + signature + MP4_DATA\n file = create_test_file(self.test_image_path, self.user, content)\n path = extract_embedded_media(file.path, file.hash)\n expected = f\"{settings.MEDIA_ROOT}/embedded_media/{file.hash}_1.mp4\"\n self.assertEqual(path, expected)\n with open(path, \"rb\") as f:\n contents = f.read()\n self.assertEqual(MP4_PREFIX + signature + MP4_DATA, contents)\n\n def test_extract_embedded_media_from_samsung_motion_photo(self):\n content = JPEG + SAMSUNG_MOTION_PHOTO_MARKER + MP4\n file = create_test_file(self.test_image_path, self.user, content)\n path = extract_embedded_media(file.path, file.hash)\n expected = f\"{settings.MEDIA_ROOT}/embedded_media/{file.hash}_1.mp4\"\n self.assertEqual(expected, path)\n with open(path, \"rb+\") as f:\n contents = f.read()\n self.assertEqual(MP4, contents)\n\n def test_fetch_embedded_media_as_owner(self):\n self.client.force_authenticate(user=self.user)\n embedded_media = create_test_file(self.test_video_path, self.user, MP4)\n photo = create_test_photo(owner=self.user)\n photo.main_file.embedded_media.add(embedded_media)\n response = self.client.get(f\"/media/embedded_media/{photo.pk}\")\n self.assertEqual(response.status_code, 200)\n\n def test_fetch_embedded_media_as_anonymous_when_photo_is_public(self):\n self.client.force_authenticate(user=None)\n embedded_media = create_test_file(self.test_video_path, self.user, MP4)\n photo = create_test_photo(owner=self.user, public=True)\n photo.main_file.embedded_media.add(embedded_media)\n\n response = self.client.get(f\"/media/embedded_media/{photo.pk}\")\n self.assertEqual(response.status_code, 200)\n\n def test_fetch_embedded_media_as_anonymous_when_photo_is_private(self):\n self.client.force_authenticate(user=None)\n embedded_media = create_test_file(self.test_video_path, self.user, MP4)\n photo = create_test_photo(owner=self.user, public=False)\n photo.main_file.embedded_media.add(embedded_media)\n\n response = self.client.get(f\"/media/embedded_media/{photo.pk}\")\n self.assertEqual(response.status_code, 404)\n\n def test_fetch_embedded_media_when_photo_does_not_have_embedded_media(self):\n self.client.force_authenticate(user=self.user)\n photo = create_test_photo(owner=self.user)\n\n response = self.client.get(f\"/media/embedded_media/{photo.pk}\")\n self.assertEqual(response.status_code, 404)\n" }, { "path": "api/filters.py", "content": "import datetime\nimport operator\nfrom functools import reduce\n\nfrom django.db.models import Q\nfrom rest_framework import filters\n\nfrom api import util\nfrom api.image_similarity import search_similar_embedding\nfrom api.semantic_search import calculate_query_embeddings\n\n\nclass SemanticSearchFilter(filters.SearchFilter):\n def filter_queryset(self, request, queryset, view):\n search_fields = self.get_search_fields(view, request)\n search_terms = self.get_search_terms(request)\n\n if not search_fields or not search_terms:\n return queryset\n\n orm_lookups = [\n self.construct_search(str(search_field), queryset=queryset)\n for search_field in search_fields\n ]\n\n if request.user.semantic_search_topk > 0:\n query = request.query_params.get(\"search\")\n start = datetime.datetime.now()\n emb, magnitude = calculate_query_embeddings(query)\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\n \"finished calculating query embedding - took %.2f seconds\" % (elapsed)\n )\n start = datetime.datetime.now()\n image_hashes = search_similar_embedding(\n request.user.id, emb, request.user.semantic_search_topk, threshold=27\n )\n elapsed = (datetime.datetime.now() - start).total_seconds()\n util.logger.info(\"search similar embedding - took %.2f seconds\" % (elapsed))\n conditions = []\n for search_term in search_terms:\n queries = [Q(**{orm_lookup: search_term}) for orm_lookup in orm_lookups]\n\n if request.user.semantic_search_topk > 0:\n queries += [Q(image_hash__in=image_hashes)]\n\n conditions.append(reduce(operator.or_, queries))\n queryset = queryset.filter(reduce(operator.and_, conditions))\n\n if self.must_call_distinct(queryset, search_fields):\n # Filtering against a many-to-many field requires us to\n # call queryset.distinct() in order to avoid duplicate items\n # in the resulting queryset.\n # We try to avoid this if possible, for performance reasons.\n queryset = queryset.distinct()\n return queryset\n" }, { "path": "api/geocode/__init__.py", "content": "GEOCODE_VERSION = \"1\"\n" }, { "path": "api/geocode/config.py", "content": "from constance import config as settings\n\nfrom .parsers.mapbox import parse as parse_mapbox\nfrom .parsers.nominatim import parse as parse_nominatim\nfrom .parsers.opencage import parse as parse_opencage\nfrom .parsers.tomtom import parse as parse_tomtom\n\n\ndef _get_config():\n return {\n \"mapbox\": {\n \"geocode_args\": {\"api_key\": settings.MAP_API_KEY},\n \"parser\": parse_mapbox,\n },\n \"maptiler\": {\n \"geocode_args\": {\"api_key\": settings.MAP_API_KEY},\n \"parser\": parse_mapbox,\n },\n \"tomtom\": {\n \"geocode_args\": {\"api_key\": settings.MAP_API_KEY},\n \"parser\": parse_tomtom,\n },\n \"nominatim\": {\n \"geocode_args\": {\"user_agent\": \"librephotos\"},\n \"parser\": parse_nominatim,\n },\n \"opencage\": {\n \"geocode_args\": {\n \"api_key\": settings.MAP_API_KEY,\n },\n \"parser\": parse_opencage,\n },\n }\n\n\ndef get_provider_config(provider) -> dict:\n config = _get_config()\n if provider not in config:\n raise Exception(f\"Map provider not found: {provider}.\")\n return config[provider][\"geocode_args\"]\n\n\ndef get_provider_parser(provider) -> callable:\n config = _get_config()\n if provider not in config:\n raise Exception(f\"Map provider not found: {provider}.\")\n return config[provider][\"parser\"]\n" }, { "path": "api/geocode/geocode.py", "content": "from typing import List\n\nimport geopy\nfrom constance import config as site_config\n\nfrom api import util\n\nfrom .config import get_provider_config, get_provider_parser\n\n\nclass Geocode:\n def __init__(self, provider):\n self._provider_config = get_provider_config(provider)\n self._parser = get_provider_parser(provider)\n self._geocoder = geopy.get_geocoder_for_service(provider)(\n **self._provider_config\n )\n\n def reverse(self, lat: float, lon: float) -> dict:\n if (\n \"geocode_args\" in self._provider_config\n and \"api_key\" in self._provider_config[\"geocode_args\"]\n and self._provider_config[\"geocode_args\"][\"api_key\"] is None\n ):\n util.logger.warning(\n \"No API key found for map provider. Please set MAP_API_KEY in the admin panel or switch map provider.\"\n )\n return {}\n location = self._geocoder.reverse(f\"{lat},{lon}\")\n return self._parser(location)\n\n def search(self, query: str, limit: int = 5) -> List[dict]:\n \"\"\"Forward geocoding: search for locations by name/address.\"\"\"\n if (\n \"api_key\" in self._provider_config\n and self._provider_config[\"api_key\"] is None\n ):\n util.logger.warning(\n \"No API key found for map provider. Please set MAP_API_KEY in the admin panel or switch map provider.\"\n )\n return []\n locations = self._geocoder.geocode(query, exactly_one=False, limit=limit)\n if not locations:\n return []\n return [\n {\n \"display_name\": loc.address,\n \"lat\": loc.latitude,\n \"lon\": loc.longitude,\n }\n for loc in locations\n ]\n\n\ndef reverse_geocode(lat: float, lon: float) -> dict:\n try:\n return Geocode(site_config.MAP_API_PROVIDER).reverse(lat, lon)\n except Exception as e:\n util.logger.warning(f\"Error while reverse geocoding: {e}\")\n return {}\n\n\ndef search_location(query: str, limit: int = 5) -> List[dict]:\n \"\"\"Search for locations by name/address using the configured map provider.\"\"\"\n try:\n return Geocode(site_config.MAP_API_PROVIDER).search(query, limit)\n except Exception as e:\n util.logger.warning(f\"Error while searching location: {e}\")\n return []\n" }, { "path": "api/geocode/parsers/__init__.py", "content": "" }, { "path": "api/geocode/parsers/mapbox.py", "content": "from api.geocode import GEOCODE_VERSION\n\n\ndef parse(location):\n context = location.raw[\"context\"]\n center = [location.raw[\"center\"][1], location.raw[\"center\"][0]]\n local_name = location.raw[\"text\"]\n places = [local_name] + [\n i[\"text\"] for i in context if not i[\"id\"].startswith(\"post\")\n ]\n return {\n \"features\": [{\"text\": place, \"center\": center} for place in places],\n \"places\": places,\n \"address\": location.address,\n \"center\": center,\n \"_v\": GEOCODE_VERSION,\n }\n" }, { "path": "api/geocode/parsers/nominatim.py", "content": "from api.geocode import GEOCODE_VERSION\n\n\ndef parse(location):\n data = location.raw[\"address\"]\n props = [\n \"road\",\n \"town\",\n \"neighbourhood\",\n \"suburb\",\n \"hamlet\",\n \"borough\",\n \"city\",\n \"county\",\n \"state\",\n \"country\",\n ]\n places = [data[prop] for prop in props if prop in data]\n center = [float(location.raw[\"lat\"]), float(location.raw[\"lon\"])]\n return {\n \"features\": [{\"text\": place, \"center\": center} for place in places],\n \"places\": places,\n \"address\": location.address,\n \"center\": center,\n \"_v\": GEOCODE_VERSION,\n }\n" }, { "path": "api/geocode/parsers/opencage.py", "content": "from api.geocode import GEOCODE_VERSION\n\n\ndef parse(location):\n data = location.raw[\"components\"]\n center = [location.raw[\"geometry\"][\"lat\"], location.raw[\"geometry\"][\"lng\"]]\n props = [\n data[\"_type\"],\n \"road\",\n \"suburb\",\n \"municipality\",\n \"hamlet\",\n \"towncity\",\n \"borough\",\n \"state\",\n \"county\",\n \"country\",\n ]\n places = [data[prop] for prop in props if prop in data]\n return {\n \"features\": [{\"text\": place, \"center\": center} for place in places],\n \"places\": places,\n \"address\": location.address,\n \"center\": center,\n \"_v\": GEOCODE_VERSION,\n }\n" }, { "path": "api/geocode/parsers/tomtom.py", "content": "from functools import reduce\n\nfrom api.geocode import GEOCODE_VERSION\n\n\ndef _dedup(iterable):\n unique_items = set()\n\n def reducer(acc, item):\n if item not in unique_items:\n unique_items.add(item)\n acc.append(item)\n return acc\n\n return reduce(reducer, iterable, [])\n\n\ndef parse(location):\n data = location.raw[\"address\"]\n address = location.address\n center = list(map(lambda x: float(x), location.raw[\"position\"].split(\",\")))\n props = [\n \"street\",\n \"streetName\",\n \"municipalitySubdivision\",\n \"countrySubdivision\",\n \"countrySecondarySubdivision\",\n \"municipality\",\n \"municipalitySubdivision\",\n \"country\",\n ]\n places = _dedup(\n [data[prop] for prop in props if prop in data and len(data[prop]) > 2]\n )\n return {\n \"features\": [{\"text\": place, \"center\": center} for place in places],\n \"places\": places,\n \"address\": address,\n \"center\": center,\n \"_v\": GEOCODE_VERSION,\n }\n" }, { "path": "api/image_captioning.py", "content": "import requests\nfrom constance import config as site_config\n\n\ndef generate_caption(image_path, blip=False, prompt=None):\n # Check if Moondream is selected as captioning model\n if site_config.CAPTIONING_MODEL == \"moondream\":\n # Use custom prompt if provided, otherwise use default caption prompt\n if prompt is None:\n prompt = \"Describe this image in a short, concise caption.\"\n\n json_data = {\n \"image_path\": image_path,\n \"prompt\": prompt,\n \"max_tokens\": 256,\n }\n try:\n response = requests.post(\"http://localhost:8008/generate\", json=json_data)\n\n if response.status_code != 201:\n print(\n f\"Error with Moondream captioning service: HTTP {response.status_code} - {response.text}\"\n )\n return \"Error generating caption with Moondream: Service unavailable\"\n\n response_data = response.json()\n return response_data[\"response\"]\n except requests.exceptions.ConnectionError:\n print(\n \"Error with Moondream captioning service: Cannot connect to LLM service on port 8008\"\n )\n return \"Error generating caption with Moondream: Service unavailable\"\n except requests.exceptions.Timeout:\n print(\"Error with Moondream captioning service: Request timeout\")\n return \"Error generating caption with Moondream: Request timeout\"\n except Exception as e:\n print(f\"Error with Moondream captioning service: {e}\")\n return \"Error generating caption with Moondream\"\n\n # Original implementation for other models\n json_data = {\n \"image_path\": image_path,\n \"onnx\": False,\n \"blip\": blip,\n }\n caption_response = requests.post(\n \"http://localhost:8007/generate-caption\", json=json_data\n ).json()\n\n return caption_response[\"caption\"]\n\n\ndef unload_model():\n requests.get(\"http://localhost:8007/unload-model\")\n" }, { "path": "api/image_similarity.py", "content": "from datetime import datetime\n\nimport numpy as np\nimport requests\nfrom django.conf import settings\nfrom django.core.paginator import Paginator\nfrom django.db.models import Q\n\nfrom api.models import Photo\nfrom api.util import logger\n\n\ndef search_similar_embedding(user, emb, result_count=100, threshold=27):\n if isinstance(user, int):\n user_id = user\n else:\n user_id = user.id\n\n image_embedding = np.array(emb, dtype=np.float32)\n\n post_data = {\n \"user_id\": user_id,\n \"image_embedding\": image_embedding.tolist(),\n \"n\": result_count,\n \"threshold\": threshold,\n }\n res = requests.post(settings.IMAGE_SIMILARITY_SERVER + \"/search/\", json=post_data)\n if res.status_code == 200:\n return res.json()[\"result\"]\n else:\n logger.error(f\"error retrieving similar embeddings for user {user_id}\")\n return []\n\n\ndef search_similar_image(user, photo, threshold=27):\n if isinstance(user, int):\n user_id = user\n else:\n user_id = user.id\n\n clip_embeddings = photo.get_clip_embeddings()\n if clip_embeddings is None:\n return []\n\n image_embedding = np.array(clip_embeddings, dtype=np.float32)\n\n post_data = {\n \"user_id\": user_id,\n \"image_embedding\": image_embedding.tolist(),\n \"threshold\": threshold,\n }\n res = requests.post(settings.IMAGE_SIMILARITY_SERVER + \"/search/\", json=post_data)\n if res.status_code == 200:\n return res.json()\n else:\n logger.error(\n f\"error retrieving similar photos to {photo.image_hash} belonging to user {user.username}\"\n )\n return []\n\n\ndef build_image_similarity_index(user):\n logger.info(f\"building similarity index for user {user.username}\")\n requests.delete(\n settings.IMAGE_SIMILARITY_SERVER + \"/build/\",\n json={\"user_id\": user.id},\n )\n start = datetime.now()\n photos = (\n Photo.objects.filter(Q(hidden=False) & Q(owner=user))\n .exclude(clip_embeddings=None)\n .only(\"clip_embeddings\", \"image_hash\")\n .order_by(\"image_hash\")\n .all()\n )\n paginator = Paginator(photos, 5000)\n\n for page in range(1, paginator.num_pages + 1):\n image_hashes = []\n image_embeddings = []\n for photo in paginator.page(page).object_list:\n clip_embeddings = photo.get_clip_embeddings()\n if clip_embeddings is not None:\n image_hashes.append(photo.image_hash)\n image_embedding = np.array(clip_embeddings, dtype=np.float32)\n image_embeddings.append(image_embedding.tolist())\n\n post_data = {\n \"user_id\": user.id,\n \"image_hashes\": image_hashes,\n \"image_embeddings\": image_embeddings,\n }\n requests.post(settings.IMAGE_SIMILARITY_SERVER + \"/build/\", json=post_data)\n elapsed = (datetime.now() - start).total_seconds()\n logger.info(\"building similarity index took %.2f seconds\" % elapsed)\n" }, { "path": "api/llm.py", "content": "import requests\nimport base64\nimport io\nfrom PIL import Image\nfrom constance import config as site_config\n\n\ndef image_to_base64_data_uri(image_path):\n \"\"\"Convert image file to base64 data URI, converting to JPEG for compatibility\"\"\"\n try:\n # Open image with PIL and convert to RGB (handles WebP, PNG with transparency, etc.)\n with Image.open(image_path) as img:\n # Convert to RGB mode (removes alpha channel if present)\n if img.mode != \"RGB\":\n img = img.convert(\"RGB\")\n\n # Save as JPEG to memory buffer\n buffer = io.BytesIO()\n img.save(buffer, format=\"JPEG\", quality=95)\n buffer.seek(0)\n\n # Encode to base64\n image_data = base64.b64encode(buffer.getvalue()).decode(\"utf-8\")\n\n return f\"data:image/jpeg;base64,{image_data}\"\n except Exception as e:\n print(f\"Error converting image to data URI: {str(e)}\")\n raise\n\n\ndef generate_prompt(prompt, image_path=None):\n if site_config.LLM_MODEL == \"none\":\n return None\n\n # Use the unified LLM service for all models including Moondream\n if site_config.LLM_MODEL == \"moondream\":\n model_path = \"/protected_media/data_models/moondream2-text-model-f16.gguf\"\n elif site_config.LLM_MODEL == \"mistral-7b-instruct-v0.2.Q5_K_M\":\n model_path = \"/protected_media/data_models/mistral-7b-instruct-v0.2.Q5_K_M.gguf\"\n else:\n return None\n\n json_data = {\n \"model_path\": model_path,\n \"max_tokens\": 64,\n \"prompt\": prompt,\n }\n\n # Convert image to base64 data URI if image path is provided\n if image_path:\n try:\n image_data = image_to_base64_data_uri(image_path)\n json_data[\"image_data\"] = image_data\n json_data[\"multimodal\"] = True\n except Exception as e:\n print(f\"Error converting image: {e}\")\n return None\n\n try:\n response = requests.post(\"http://localhost:8008/generate\", json=json_data)\n\n if response.status_code != 201:\n print(\n f\"Error with LLM service: HTTP {response.status_code} - {response.text}\"\n )\n return None\n\n response_data = response.json()\n return response_data.get(\"response\", \"\")\n except requests.exceptions.ConnectionError:\n print(\"Error with LLM service: Cannot connect to service on port 8008\")\n return None\n except requests.exceptions.Timeout:\n print(\"Error with LLM service: Request timeout\")\n return None\n except Exception as e:\n print(f\"Error with LLM service: {e}\")\n return None\n" }, { "path": "api/management/__init__.py", "content": "" }, { "path": "api/management/commands/build_similarity_index.py", "content": "from django.core.management.base import BaseCommand\nfrom django_q.tasks import AsyncTask\n\nfrom api.image_similarity import build_image_similarity_index\nfrom api.models import User\n\n\nclass Command(BaseCommand):\n help = \"Build image similarity index for all users\"\n\n def handle(self, *args, **kwargs):\n for user in User.objects.all():\n AsyncTask(build_image_similarity_index, user).run()\n" }, { "path": "api/management/commands/clear_cache.py", "content": "from django.conf import settings\nfrom django.core.cache import cache\nfrom django.core.management.base import BaseCommand, CommandError\n\n\nclass Command(BaseCommand):\n \"\"\"A simple management command which clears the site-wide cache.\"\"\"\n\n help = \"Fully clear your site-wide cache.\"\n\n def handle(self, *args, **kwargs):\n try:\n assert settings.CACHES\n cache.clear()\n self.stdout.write(\"Your cache has been cleared!\\n\")\n except AttributeError:\n raise CommandError(\"You have no cache configured!\\n\")\n" }, { "path": "api/management/commands/createadmin.py", "content": "import os\nimport sys\n\nfrom django.core.management.base import BaseCommand, CommandError\nfrom django.core.validators import ValidationError, validate_email\n\nfrom api.models import User\n\n\nclass Command(BaseCommand):\n help = \"Create a LibrePhotos user with administrative permissions\"\n\n def add_arguments(self, parser):\n parser.add_argument(\"admin_username\", help=\"Username to create\")\n parser.add_argument(\"admin_email\", help=\"Email address of the new user\")\n parser.add_argument(\n \"-u\",\n \"--update\",\n help=(\n \"Update an existing superuser's password (ignoring the\"\n \"provided email) instead of reporting an error\"\n ),\n action=\"store_true\",\n )\n # Done this way because command lines are visible to the whole system by\n # default on Linux, so a password in the arguments would leak\n parser.epilog = (\n \"The password is read from the ADMIN_PASSWORD\"\n \"environment variable or interactively if\"\n \"ADMIN_PASSWORD is not set\"\n )\n\n def handle(self, *args, **options):\n try:\n validate_email(options[\"admin_email\"])\n except ValidationError as err:\n raise CommandError(err.message)\n\n if \"ADMIN_PASSWORD\" in os.environ:\n options[\"admin_password\"] = os.environ[\"ADMIN_PASSWORD\"]\n else:\n options[\"admin_password\"] = User.objects.make_random_password()\n\n if not options[\"admin_password\"]:\n raise CommandError(\"Admin password cannot be empty\")\n\n if not User.objects.filter(username=options[\"admin_username\"].lower()).exists():\n User.objects.create_superuser(\n options[\"admin_username\"].lower(),\n options[\"admin_email\"],\n options[\"admin_password\"],\n )\n elif options[\"update\"]:\n print(\n \"Warning: ignoring provided email \" + options[\"admin_email\"],\n file=sys.stderr,\n )\n admin_user = User.objects.get(username=options[\"admin_username\"].lower())\n admin_user.set_password(options[\"admin_password\"])\n admin_user.save()\n else:\n raise CommandError(\"Specified user already exists\")\n" }, { "path": "api/management/commands/createuser.py", "content": "import os\nimport sys\n\nfrom django.core.management.base import BaseCommand, CommandError\nfrom django.core.validators import ValidationError, validate_email\n\nfrom api.models import User\n\n\nclass Command(BaseCommand):\n help = \"Create a LibrePhotos user\"\n\n def add_arguments(self, parser):\n parser.add_argument(\"username\", help=\"Username to create\")\n parser.add_argument(\"email\", help=\"Email address of the new user\")\n parser.add_argument(\n \"--password\",\n help=\"Password to create/update for user. (autogenerate if omitted)\",\n )\n parser.add_argument(\n \"--update\",\n help=(\n \"Update an existing user's password (ignoring the provided email) \"\n \"instead of reporting an error\"\n ),\n action=\"store_true\",\n )\n parser.add_argument(\n \"--admin\",\n help=\"Create user with administrative privileges\",\n action=\"store_true\",\n )\n # Done this way because command lines are visible to the whole system by\n # default on Linux, so a password in the arguments would leak\n parser.epilog = (\n \"When creating user with administrative privileges,\"\n \"the password is read from the ADMIN_PASSWORD\"\n \"environment variable or interactively if\"\n \"ADMIN_PASSWORD is not set\"\n )\n\n def handle(self, *args, **options):\n try:\n validate_email(options[\"email\"])\n except ValidationError as err:\n raise CommandError(err.message)\n\n if options[\"admin\"] and \"ADMIN_PASSWORD\" in os.environ:\n options[\"password\"] = os.environ[\"ADMIN_PASSWORD\"]\n\n if not options[\"password\"]:\n options[\"password\"] = User.objects.make_random_password()\n\n if not User.objects.filter(username=options[\"username\"].lower()).exists():\n if options[\"admin\"]:\n User.objects.create_superuser(\n options[\"username\"].lower(),\n options[\"email\"],\n options[\"password\"],\n )\n else:\n User.objects.create_user(\n options[\"username\"].lower(),\n options[\"email\"],\n options[\"password\"],\n )\n\n elif options[\"update\"]:\n print(\n \"Warning: ignoring provided email \" + options[\"email\"],\n file=sys.stderr,\n )\n user = User.objects.get(username=options[\"username\"].lower())\n user.set_password(options[\"password\"])\n user.save()\n else:\n raise CommandError(\"Specified user already exists\")\n" }, { "path": "api/management/commands/save_metadata.py", "content": "from django.core.management.base import BaseCommand\n\nfrom api.models import Photo, User\nfrom api.models.person import Person\n\n\nclass Command(BaseCommand):\n help = \"Save metadata to image files (or XMP sidecar files)\"\n\n def add_arguments(self, parser):\n parser.add_argument(\n \"--types\",\n nargs=\"+\",\n choices=[\"ratings\", \"face_tags\"],\n default=[\"ratings\"],\n help=\"Which metadata types to write (default: ratings)\",\n )\n parser.add_argument(\n \"--user\",\n type=str,\n help=\"Only process photos owned by this username\",\n )\n parser.add_argument(\n \"--sidecar\",\n action=\"store_true\",\n default=True,\n help=\"Write to XMP sidecar files (default)\",\n )\n parser.add_argument(\n \"--media-file\",\n action=\"store_true\",\n help=\"Write directly to media files instead of sidecars\",\n )\n parser.add_argument(\n \"--dry-run\",\n action=\"store_true\",\n help=\"Only show what would be written, don't actually write\",\n )\n\n def handle(self, *args, **options):\n metadata_types = options[\"types\"]\n use_sidecar = not options[\"media_file\"]\n\n photos = Photo.objects.all()\n\n if options[\"user\"]:\n try:\n user = User.objects.get(username=options[\"user\"])\n photos = photos.filter(owner=user)\n except User.DoesNotExist:\n self.stderr.write(f\"User '{options['user']}' not found\")\n return\n\n # When only writing face tags, filter to photos with any (non-deleted) faces\n if metadata_types == [\"face_tags\"]:\n photos = photos.filter(\n faces__deleted=False,\n ).distinct()\n\n total = photos.count()\n self.stdout.write(f\"Found {total} photos to process (types: {metadata_types})\")\n\n if options[\"dry_run\"]:\n self.stdout.write(\"Dry run — no files will be modified\")\n return\n\n written = 0\n errors = 0\n for i, photo in enumerate(photos.iterator(), 1):\n try:\n photo._save_metadata(\n use_sidecar=use_sidecar, metadata_types=metadata_types\n )\n written += 1\n except Exception as e:\n errors += 1\n self.stderr.write(f\"Error writing {photo.image_hash}: {e}\")\n\n if i % 100 == 0:\n self.stdout.write(\n f\"Progress: {i}/{total} ({written} written, {errors} errors)\"\n )\n\n self.stdout.write(\n self.style.SUCCESS(\n f\"Done. {written} written, {errors} errors out of {total} photos.\"\n )\n )\n" }, { "path": "api/management/commands/scan.py", "content": "import traceback\nimport uuid\n\nfrom django.core.management.base import BaseCommand\n\nfrom api.directory_watcher import scan_photos\nfrom api.models import User\nfrom api.models.user import get_deleted_user\nfrom nextcloud.directory_watcher import scan_photos as scan_photos_nextcloud\n\n\nclass Command(BaseCommand):\n help = \"scan directory for all users\"\n\n def add_arguments(self, parser):\n parser_group = parser.add_mutually_exclusive_group()\n parser_group.add_argument(\n \"-f\", \"--full-scan\", help=(\"Run full directory scan\"), action=\"store_true\"\n )\n parser_group.add_argument(\n \"-s\", \"--scan-files\", help=(\"Scan a list of files\"), nargs=\"+\", default=[]\n )\n parser_group.add_argument(\n \"-n\",\n \"--nextcloud\",\n help=(\"Run nextcloud scan instead of directory scan\"),\n action=\"store_true\",\n )\n\n def handle(self, *args, **options):\n # Nextcloud scan\n if options[\"nextcloud\"]:\n self.nextcloud_scan()\n return\n\n # Add a single file.\n if options[\"scan_files\"]:\n scan_files = options[\"scan_files\"]\n deleted_user: User = get_deleted_user()\n for user in User.objects.all():\n user_files = []\n if user == deleted_user:\n continue\n for scan_file in scan_files:\n if scan_file.startswith(user.scan_directory):\n user_files.append(scan_file)\n if user_files:\n scan_photos(user, False, uuid.uuid4(), scan_files=user_files)\n return\n\n # Directory scan\n deleted_user: User = get_deleted_user()\n for user in User.objects.all():\n if user != deleted_user:\n scan_photos(\n user, options[\"full_scan\"], uuid.uuid4(), user.scan_directory\n )\n\n def nextcloud_scan(self):\n for user in User.objects.all():\n if not user.nextcloud_scan_directory:\n print(\n f\"Skipping nextcloud scan for user {user.username}. No scan directory configured.\"\n )\n continue\n print(f\"Starting nextcloud scan for user {user.username}.\")\n try:\n scan_photos_nextcloud(user, uuid.uuid4())\n except Exception:\n print(f\"Nextcloud scan for user {user.username} failed:\")\n print(traceback.format_exc())\n" }, { "path": "api/management/commands/start_cleaning_service.py", "content": "from django.core.management.base import BaseCommand\nfrom django_q.models import Schedule\nfrom django_q.tasks import schedule\n\nfrom api.util import logger\n\n\nclass Command(BaseCommand):\n help = \"Start the cleanup service.\"\n\n def handle(self, *args, **kwargs):\n if not Schedule.objects.filter(\n func=\"api.services.cleanup_deleted_photos\"\n ).exists():\n schedule(\n \"api.services.cleanup_deleted_photos\",\n schedule_type=Schedule.DAILY,\n )\n logger.info(\"Cleanup service started\")\n" }, { "path": "api/management/commands/start_job_cleanup_service.py", "content": "from django.core.management.base import BaseCommand\nfrom django_q.models import Schedule\nfrom django_q.tasks import schedule\n\nfrom api.util import logger\n\n\nclass Command(BaseCommand):\n help = \"Start the job cleanup service to mark stuck jobs as failed and remove old completed jobs.\"\n\n def handle(self, *args, **kwargs):\n # Schedule hourly cleanup of stuck jobs (running for more than 24 hours)\n if not Schedule.objects.filter(\n func=\"api.models.long_running_job.LongRunningJob.cleanup_stuck_jobs\"\n ).exists():\n schedule(\n \"api.models.long_running_job.LongRunningJob.cleanup_stuck_jobs\",\n schedule_type=Schedule.HOURLY,\n )\n logger.info(\"Scheduled hourly stuck job cleanup\")\n\n # Schedule daily cleanup of old completed jobs (older than 30 days)\n if not Schedule.objects.filter(\n func=\"api.models.long_running_job.LongRunningJob.cleanup_old_jobs\"\n ).exists():\n schedule(\n \"api.models.long_running_job.LongRunningJob.cleanup_old_jobs\",\n schedule_type=Schedule.DAILY,\n )\n logger.info(\"Scheduled daily old job cleanup\")\n\n logger.info(\"Job cleanup service started\")\n" }, { "path": "api/management/commands/start_service.py", "content": "from django.core.management.base import BaseCommand\nfrom django_q.models import Schedule\nfrom django_q.tasks import schedule\n\nfrom api.services import SERVICES, start_service\n\n\nclass Command(BaseCommand):\n help = \"Start one of the services.\"\n\n # Define all the services that can be started\n def add_arguments(self, parser):\n parser.add_argument(\n \"service\",\n type=str,\n help=\"The service to start\",\n choices=[\n SERVICES.keys(),\n \"all\",\n ],\n )\n\n def handle(self, *args, **kwargs):\n service = kwargs[\"service\"]\n if service == \"all\":\n for svc in SERVICES.keys():\n start_service(svc)\n if not Schedule.objects.filter(func=\"api.services.check_services\").exists():\n schedule(\n \"api.services.check_services\",\n schedule_type=Schedule.MINUTES,\n minutes=1,\n )\n else:\n start_service(service)\n" }, { "path": "api/metadata/__init__.py", "content": "# api/metadata — organized metadata reading, writing, and tag constants.\n#\n# Submodules:\n# api.metadata.tags — Tag name constants (Tags class)\n# api.metadata.reader — get_metadata(), sidecar file helpers\n# api.metadata.writer — write_metadata()\n# api.metadata.face_regions — face region coordinate conversion & tag building\n#\n# Import directly from submodules to avoid circular import issues:\n# from api.metadata.tags import Tags\n# from api.metadata.reader import get_metadata\n# from api.metadata.writer import write_metadata\n# from api.metadata.face_regions import get_face_region_tags\n" }, { "path": "api/metadata/face_regions.py", "content": "import PIL\n\nfrom api.metadata.reader import get_metadata\nfrom api.metadata.tags import Tags\nfrom api.models.face import Face\nfrom api.models.person import Person\nfrom api.util import logger\n\n\ndef thumbnail_coords_to_normalized(top, right, bottom, left, thumb_width, thumb_height):\n \"\"\"Convert Face model pixel coords (in big thumbnail space) to MWG-RS\n normalized center-based coords.\"\"\"\n center_x = (left + right) / 2.0 / thumb_width\n center_y = (top + bottom) / 2.0 / thumb_height\n w = (right - left) / thumb_width\n h = (bottom - top) / thumb_height\n return center_x, center_y, w, h\n\n\ndef reverse_orientation_transform(x, y, w, h, orientation):\n \"\"\"Invert the orientation transforms from face_extractor.py lines 54-80.\n\n The read path applies a forward transform from XMP coords to display coords.\n This function reverses that so we can go from display coords back to XMP coords.\n \"\"\"\n if orientation == \"Rotate 90 CW\":\n # Forward: x' = 1 - y, y' = x, swap w/h\n # Reverse: y_orig = 1 - x', x_orig = y'\n new_x = y\n new_y = 1 - x\n w, h = h, w\n return new_x, new_y, w, h\n elif orientation == \"Mirror horizontal\":\n # Forward: x' = 1 - x\n # Reverse: x = 1 - x'\n return 1 - x, y, w, h\n elif orientation == \"Rotate 180\":\n # Forward: x' = 1 - x, y' = 1 - y\n # Reverse: same\n return 1 - x, 1 - y, w, h\n elif orientation == \"Mirror vertical\":\n # Forward: y' = 1 - y\n # Reverse: y = 1 - y'\n return x, 1 - y, w, h\n elif orientation == \"Mirror horizontal and rotate 270 CW\":\n # Forward: x' = 1 - y, y' = x, swap w/h\n # Same as Rotate 90 CW (the mirror cancels differently)\n new_x = y\n new_y = 1 - x\n w, h = h, w\n return new_x, new_y, w, h\n elif orientation == \"Mirror horizontal and rotate 90 CW\":\n # Forward: x' = y, y' = 1 - x, swap w/h\n # Reverse: x_orig = 1 - y', y_orig = x'\n new_x = 1 - y\n new_y = x\n w, h = h, w\n return new_x, new_y, w, h\n elif orientation == \"Rotate 270 CW\":\n # Forward: x' = y, y' = 1 - x, swap w/h\n # Reverse: x_orig = 1 - y', y_orig = x'\n new_x = 1 - y\n new_y = x\n w, h = h, w\n return new_x, new_y, w, h\n # Normal orientation or unknown — no transform\n return x, y, w, h\n\n\ndef _escape_exiftool_value(value):\n \"\"\"Escape special characters in a string value for exiftool structured data.\n\n ExifTool uses commas, equals, braces in its structured value syntax,\n so person names containing these characters need escaping.\n \"\"\"\n # ExifTool expects special chars to be escaped with backslash\n for ch in (\"\\\\\", \"{\", \"}\", \"=\", \",\"):\n value = value.replace(ch, f\"\\\\{ch}\")\n return value\n\n\ndef build_face_region_exiftool_args(face_regions, image_width=None, image_height=None):\n \"\"\"Build exiftool args dict for writing XMP-mwg-rs:RegionInfo and XMP:Subject.\n\n Args:\n face_regions: list of dicts with keys: name, x, y, w, h\n image_width: original image width in pixels (for AppliedToDimensions)\n image_height: original image height in pixels (for AppliedToDimensions)\n\n Returns:\n dict of tag -> value suitable for write_metadata()\n \"\"\"\n region_parts = []\n person_names = []\n for region in face_regions:\n name = _escape_exiftool_value(region[\"name\"])\n x = f\"{region['x']:.6f}\"\n y = f\"{region['y']:.6f}\"\n w = f\"{region['w']:.6f}\"\n h = f\"{region['h']:.6f}\"\n region_parts.append(\n f\"{{Area={{X={x},Y={y},W={w},H={h},Unit=normalized}}\"\n f\",Name={name},Type=Face}}\"\n )\n if region[\"name\"]:\n person_names.append(region[\"name\"])\n\n region_list = \",\".join(region_parts)\n\n # Include AppliedToDimensions if image dimensions are available\n if image_width and image_height:\n applied_to = f\"AppliedToDimensions={{W={image_width},H={image_height},Unit=pixel}},\"\n else:\n applied_to = \"\"\n\n value = f\"{{{applied_to}RegionList=[{region_list}]}}\"\n\n tags = {Tags.REGION_INFO_WRITE: value}\n\n # Add person names as XMP:Subject keywords for Lightroom compatibility\n if person_names:\n tags[Tags.SUBJECT] = person_names\n\n return tags\n\n\ndef get_face_region_tags(photo):\n \"\"\"Build face region exiftool tags dict for a photo.\n\n Returns a dict of tags suitable for merging into _save_metadata()'s tags_to_write,\n or an empty dict if no faces exist.\n\n Args:\n photo: Photo model instance\n\n Returns:\n dict: e.g. {\"XMP-mwg-rs:RegionInfo\": \"{RegionList=[...]}\"} or {}\n \"\"\"\n faces = Face.objects.filter(\n photo=photo,\n deleted=False,\n ).select_related(\"person\")\n\n if not faces.exists():\n return {}\n\n # Get thumbnail dimensions\n try:\n thumb_path = photo.thumbnail.thumbnail_big.path\n thumb_image = PIL.Image.open(thumb_path)\n thumb_width, thumb_height = thumb_image.size\n thumb_image.close()\n except Exception:\n logger.error(\n f\"Cannot open thumbnail for photo {photo.image_hash}, skipping face tags\"\n )\n return {}\n\n # Get EXIF orientation and original image dimensions\n (orientation, image_width, image_height) = get_metadata(\n photo.main_file.path,\n tags=[Tags.ORIENTATION, Tags.IMAGE_WIDTH, Tags.IMAGE_HEIGHT],\n try_sidecar=True,\n )\n\n # Convert each face's coordinates\n face_regions = []\n for face in faces:\n x, y, w, h = thumbnail_coords_to_normalized(\n face.location_top,\n face.location_right,\n face.location_bottom,\n face.location_left,\n thumb_width,\n thumb_height,\n )\n x, y, w, h = reverse_orientation_transform(x, y, w, h, orientation)\n # Only write person name for user-labeled faces\n if face.person and face.person.kind == Person.KIND_USER:\n name = face.person.name\n else:\n name = \"\"\n face_regions.append(\n {\n \"name\": name,\n \"x\": x,\n \"y\": y,\n \"w\": w,\n \"h\": h,\n }\n )\n\n return build_face_region_exiftool_args(face_regions, image_width, image_height)\n" }, { "path": "api/metadata/reader.py", "content": "import os\nimport os.path\n\nimport requests\n\n\ndef get_sidecar_files_in_priority_order(media_file):\n \"\"\"Returns a list of possible XMP sidecar files for *media_file*, ordered\n by priority.\n\n \"\"\"\n image_basename = os.path.splitext(media_file)[0]\n return [\n image_basename + \".xmp\",\n image_basename + \".XMP\",\n media_file + \".xmp\",\n media_file + \".XMP\",\n ]\n\n\ndef _get_existing_metadata_files_reversed(media_file, include_sidecar_files):\n if include_sidecar_files:\n files = [\n file\n for file in get_sidecar_files_in_priority_order(media_file)\n if os.path.exists(file)\n ]\n files.append(media_file)\n return list(reversed(files))\n return [media_file]\n\n\ndef get_metadata(media_file, tags, try_sidecar=True, struct=False):\n \"\"\"Get values for each metadata tag in *tags* from *media_file*.\n If *try_sidecar* is `True`, use the value set in any XMP sidecar file\n stored alongside *media_file*.\n If *struct* is `True`, use the exiftool instance which returns structured data\n\n Returns a list with the value of each tag in *tags* or `None` if the\n tag was not found.\n\n \"\"\"\n files_by_reverse_priority = _get_existing_metadata_files_reversed(\n media_file, try_sidecar\n )\n\n json = {\n \"tags\": tags,\n \"files_by_reverse_priority\": files_by_reverse_priority,\n \"struct\": struct,\n }\n response = requests.post(\"http://localhost:8010/get-tags\", json=json).json()\n return response[\"values\"]\n" }, { "path": "api/metadata/tags.py", "content": "class Tags:\n RATING = \"Rating\"\n IMAGE_HEIGHT = \"ImageHeight\"\n IMAGE_WIDTH = \"ImageWidth\"\n DATE_TIME_ORIGINAL = \"EXIF:DateTimeOriginal\"\n DATE_TIME = \"EXIF:DateTime\"\n QUICKTIME_CREATE_DATE = \"QuickTime:CreateDate\"\n QUICKTIME_DURATION = \"QuickTime:Duration\"\n LATITUDE = \"Composite:GPSLatitude\"\n LONGITUDE = \"Composite:GPSLongitude\"\n GPS_DATE_TIME = \"Composite:GPSDateTime\"\n FILE_SIZE = \"File:FileSize\"\n FSTOP = \"EXIF:FNumber\"\n EXPOSURE_TIME = \"EXIF:ExposureTime\"\n ISO = \"EXIF:ISOSpeedRatings\"\n FOCAL_LENGTH = \"EXIF:FocalLength\"\n FOCAL_LENGTH_35MM = \"EXIF:FocalLengthIn35mmFilm\"\n SHUTTER_SPEED = \"EXIF:ShutterSpeedValue\"\n CAMERA = \"EXIF:Model\"\n LENS = \"EXIF:LensModel\"\n SUBJECT_DISTANCE = \"EXIF:SubjectDistance\"\n DIGITAL_ZOOM_RATIO = \"EXIF:DigitalZoomRatio\"\n REGION_INFO = \"XMP:RegionInfo\"\n REGION_INFO_WRITE = \"XMP-mwg-rs:RegionInfo\"\n SUBJECT = \"XMP:Subject\"\n ROTATION = \"QuickTime:Rotation\"\n ORIENTATION = \"EXIF:Orientation\"\n\n # Burst/sequence detection tags\n SUBSEC_TIME_ORIGINAL = \"EXIF:SubSecTimeOriginal\"\n SUBSEC_TIME = \"EXIF:SubSecTime\"\n IMAGE_NUMBER = \"EXIF:ImageNumber\"\n IMAGE_UNIQUE_ID = \"EXIF:ImageUniqueID\"\n BURST_MODE = \"MakerNotes:BurstMode\"\n CONTINUOUS_DRIVE = \"MakerNotes:ContinuousDrive\"\n SEQUENCE_NUMBER = \"MakerNotes:SequenceNumber\"\n # Camera serial number (useful for grouping shots from same camera)\n SERIAL_NUMBER = \"EXIF:SerialNumber\"\n CAMERA_SERIAL = \"MakerNotes:SerialNumber\"\n" }, { "path": "api/metadata/writer.py", "content": "import os\n\nimport exiftool\n\nfrom api.metadata.reader import get_sidecar_files_in_priority_order\nfrom api.util import logger\n\n\ndef write_metadata(media_file, tags, use_sidecar=True):\n et = exiftool.ExifTool()\n terminate_et = False\n if not et.running:\n et.start()\n terminate_et = True\n # To-Do: Replace with new File Structure\n if use_sidecar:\n file_path = get_sidecar_files_in_priority_order(media_file)[0]\n else:\n file_path = media_file\n\n try:\n logger.info(f\"Writing {tags} to {file_path}\")\n params = []\n for tag, value in tags.items():\n if isinstance(value, list):\n for item in value:\n params.append(os.fsencode(f\"-{tag}={item}\"))\n else:\n params.append(os.fsencode(f\"-{tag}={value}\"))\n params.append(b\"-overwrite_original\")\n params.append(os.fsencode(file_path))\n et.execute(*params)\n finally:\n if terminate_et:\n et.terminate()\n" }, { "path": "api/middleware.py", "content": "class FingerPrintMiddleware:\n def __init__(self, get_response):\n self.get_response = get_response\n # One-time configuration and initializatio\n\n def __call__(self, request):\n response = self.get_response(request)\n import hashlib\n\n fingerprint_raw = \"\".join(\n (\n request.META.get(\"HTTP_USER_AGENT\", \"\"),\n request.META.get(\"HTTP_ACCEPT_ENCODING\", \"\"),\n )\n )\n # print(fingerprint_raw)\n fingerprint = hashlib.md5(fingerprint_raw.encode(\"utf-8\")).hexdigest()\n request.fingerprint = fingerprint\n # print(fingerprint)\n return response\n" }, { "path": "api/migrations/0001_initial.py", "content": "# Generated by Django 2.1.2 on 2020-11-15 18:49\n\nimport datetime\n\nimport django.contrib.auth.models\nimport django.contrib.auth.validators\nimport django.contrib.postgres.fields.jsonb\nimport django.utils.timezone\nimport django_cryptography.fields\nfrom django.conf import settings\nfrom django.db import migrations, models\n\nimport api.models\n\n\nclass Migration(migrations.Migration):\n initial = True\n\n dependencies = [\n (\"auth\", \"0009_alter_user_last_name_max_length\"),\n ]\n\n operations = [\n migrations.CreateModel(\n name=\"User\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"password\", models.CharField(max_length=128, verbose_name=\"password\")),\n (\n \"last_login\",\n models.DateTimeField(\n blank=True, null=True, verbose_name=\"last login\"\n ),\n ),\n (\n \"is_superuser\",\n models.BooleanField(\n default=False,\n help_text=\"Designates that this user has all permissions without explicitly assigning them.\",\n verbose_name=\"superuser status\",\n ),\n ),\n (\n \"username\",\n models.CharField(\n error_messages={\n \"unique\": \"A user with that username already exists.\"\n },\n help_text=\"Required. 150 characters or fewer. Letters, digits and @/./+/-/_ only.\",\n max_length=150,\n unique=True,\n validators=[\n django.contrib.auth.validators.UnicodeUsernameValidator()\n ],\n verbose_name=\"username\",\n ),\n ),\n (\n \"first_name\",\n models.CharField(\n blank=True, max_length=30, verbose_name=\"first name\"\n ),\n ),\n (\n \"last_name\",\n models.CharField(\n blank=True, max_length=150, verbose_name=\"last name\"\n ),\n ),\n (\n \"email\",\n models.EmailField(\n blank=True, max_length=254, verbose_name=\"email address\"\n ),\n ),\n (\n \"is_staff\",\n models.BooleanField(\n default=False,\n help_text=\"Designates whether the user can log into this admin site.\",\n verbose_name=\"staff status\",\n ),\n ),\n (\n \"is_active\",\n models.BooleanField(\n default=True,\n help_text=\"Designates whether this user should be treated as active. Unselect this instead of deleting accounts.\",\n verbose_name=\"active\",\n ),\n ),\n (\n \"date_joined\",\n models.DateTimeField(\n default=django.utils.timezone.now, verbose_name=\"date joined\"\n ),\n ),\n (\"scan_directory\", models.CharField(db_index=True, max_length=512)),\n (\"avatar\", models.ImageField(null=True, upload_to=\"avatars\")),\n (\n \"nextcloud_server_address\",\n models.CharField(default=None, max_length=200, null=True),\n ),\n (\n \"nextcloud_username\",\n models.CharField(default=None, max_length=64, null=True),\n ),\n (\n \"nextcloud_app_password\",\n django_cryptography.fields.encrypt(\n models.CharField(default=None, max_length=64, null=True)\n ),\n ),\n (\n \"nextcloud_scan_directory\",\n models.CharField(db_index=True, max_length=512, null=True),\n ),\n (\n \"groups\",\n models.ManyToManyField(\n blank=True,\n help_text=\"The groups this user belongs to. A user will get all permissions granted to each of their groups.\",\n related_name=\"user_set\",\n related_query_name=\"user\",\n to=\"auth.Group\",\n verbose_name=\"groups\",\n ),\n ),\n (\n \"user_permissions\",\n models.ManyToManyField(\n blank=True,\n help_text=\"Specific permissions for this user.\",\n related_name=\"user_set\",\n related_query_name=\"user\",\n to=\"auth.Permission\",\n verbose_name=\"user permissions\",\n ),\n ),\n ],\n options={\n \"verbose_name\": \"user\",\n \"verbose_name_plural\": \"users\",\n \"abstract\": False,\n },\n managers=[\n (\"objects\", django.contrib.auth.models.UserManager()),\n ],\n ),\n migrations.CreateModel(\n name=\"AlbumAuto\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"title\", models.CharField(blank=True, max_length=512, null=True)),\n (\"timestamp\", models.DateTimeField(db_index=True)),\n (\"created_on\", models.DateTimeField(db_index=True)),\n (\"gps_lat\", models.FloatField(blank=True, null=True)),\n (\"gps_lon\", models.FloatField(blank=True, null=True)),\n (\"favorited\", models.BooleanField(db_index=True, default=False)),\n (\n \"owner\",\n models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ],\n ),\n migrations.CreateModel(\n name=\"AlbumDate\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\n \"title\",\n models.CharField(\n blank=True, db_index=True, max_length=512, null=True\n ),\n ),\n (\"date\", models.DateField(db_index=True, null=True)),\n (\"favorited\", models.BooleanField(db_index=True, default=False)),\n (\n \"location\",\n django.contrib.postgres.fields.jsonb.JSONField(\n blank=True, db_index=True, null=True\n ),\n ),\n (\n \"owner\",\n models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ],\n ),\n migrations.CreateModel(\n name=\"AlbumPlace\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"title\", models.CharField(db_index=True, max_length=512)),\n (\"geolocation_level\", models.IntegerField(db_index=True, null=True)),\n (\"favorited\", models.BooleanField(db_index=True, default=False)),\n ],\n ),\n migrations.CreateModel(\n name=\"AlbumThing\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"title\", models.CharField(db_index=True, max_length=512)),\n (\n \"thing_type\",\n models.CharField(db_index=True, max_length=512, null=True),\n ),\n (\"favorited\", models.BooleanField(db_index=True, default=False)),\n ],\n ),\n migrations.CreateModel(\n name=\"AlbumUser\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"title\", models.CharField(max_length=512)),\n (\"created_on\", models.DateTimeField(auto_now=True, db_index=True)),\n (\"favorited\", models.BooleanField(db_index=True, default=False)),\n (\"public\", models.BooleanField(db_index=True, default=False)),\n ],\n ),\n migrations.CreateModel(\n name=\"Face\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"image\", models.ImageField(upload_to=\"faces\")),\n (\"image_path\", models.FilePathField()),\n (\"person_label_is_inferred\", models.NullBooleanField(db_index=True)),\n (\n \"person_label_probability\",\n models.FloatField(db_index=True, default=0.0),\n ),\n (\"location_top\", models.IntegerField()),\n (\"location_bottom\", models.IntegerField()),\n (\"location_left\", models.IntegerField()),\n (\"location_right\", models.IntegerField()),\n (\"encoding\", models.TextField()),\n ],\n ),\n migrations.CreateModel(\n name=\"LongRunningJob\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\n \"job_type\",\n models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n ]\n ),\n ),\n (\"finished\", models.BooleanField(default=False)),\n (\"failed\", models.BooleanField(default=False)),\n (\"job_id\", models.CharField(db_index=True, max_length=36, unique=True)),\n (\"queued_at\", models.DateTimeField(default=datetime.datetime.now)),\n (\"started_at\", models.DateTimeField(null=True)),\n (\"finished_at\", models.DateTimeField(null=True)),\n (\n \"result\",\n django.contrib.postgres.fields.jsonb.JSONField(\n default={\"progress\": {\"target\": 0, \"current\": 0}}\n ),\n ),\n (\n \"started_by\",\n models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ],\n ),\n migrations.CreateModel(\n name=\"Person\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"name\", models.CharField(max_length=128)),\n (\n \"kind\",\n models.CharField(\n choices=[\n (\"USER\", \"User Labelled\"),\n (\"CLUSTER\", \"Cluster ID\"),\n (\"UNKNOWN\", \"Unknown Person\"),\n ],\n max_length=10,\n ),\n ),\n (\"mean_face_encoding\", models.TextField()),\n (\"cluster_id\", models.IntegerField(null=True)),\n (\n \"account\",\n models.OneToOneField(\n default=None,\n null=True,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ],\n ),\n migrations.CreateModel(\n name=\"Photo\",\n fields=[\n (\"image_path\", models.CharField(db_index=True, max_length=512)),\n (\n \"image_hash\",\n models.CharField(max_length=64, primary_key=True, serialize=False),\n ),\n (\"thumbnail\", models.ImageField(upload_to=\"thumbnails\")),\n (\"thumbnail_tiny\", models.ImageField(upload_to=\"thumbnails_tiny\")),\n (\"thumbnail_small\", models.ImageField(upload_to=\"thumbnails_small\")),\n (\"thumbnail_big\", models.ImageField(upload_to=\"thumbnails_big\")),\n (\"square_thumbnail\", models.ImageField(upload_to=\"square_thumbnails\")),\n (\n \"square_thumbnail_tiny\",\n models.ImageField(upload_to=\"square_thumbnails_tiny\"),\n ),\n (\n \"square_thumbnail_small\",\n models.ImageField(upload_to=\"square_thumbnails_small\"),\n ),\n (\n \"square_thumbnail_big\",\n models.ImageField(upload_to=\"square_thumbnails_big\"),\n ),\n (\"image\", models.ImageField(upload_to=\"photos\")),\n (\"added_on\", models.DateTimeField(db_index=True)),\n (\"exif_gps_lat\", models.FloatField(blank=True, null=True)),\n (\"exif_gps_lon\", models.FloatField(blank=True, null=True)),\n (\n \"exif_timestamp\",\n models.DateTimeField(blank=True, db_index=True, null=True),\n ),\n (\n \"exif_json\",\n django.contrib.postgres.fields.jsonb.JSONField(\n blank=True, null=True\n ),\n ),\n (\n \"geolocation_json\",\n django.contrib.postgres.fields.jsonb.JSONField(\n blank=True, db_index=True, null=True\n ),\n ),\n (\n \"captions_json\",\n django.contrib.postgres.fields.jsonb.JSONField(\n blank=True, db_index=True, null=True\n ),\n ),\n (\n \"search_captions\",\n models.TextField(blank=True, db_index=True, null=True),\n ),\n (\n \"search_location\",\n models.TextField(blank=True, db_index=True, null=True),\n ),\n (\"favorited\", models.BooleanField(db_index=True, default=False)),\n (\"hidden\", models.BooleanField(db_index=True, default=False)),\n (\"public\", models.BooleanField(db_index=True, default=False)),\n (\"encoding\", models.TextField(default=None, null=True)),\n (\n \"owner\",\n models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n (\n \"shared_to\",\n models.ManyToManyField(\n related_name=\"photo_shared_to\", to=settings.AUTH_USER_MODEL\n ),\n ),\n ],\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"person\",\n field=models.ForeignKey(\n on_delete=models.SET(api.models.person.get_unknown_person),\n related_name=\"faces\",\n to=\"api.Person\",\n ),\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"photo\",\n field=models.ForeignKey(\n null=True,\n on_delete=models.SET(api.models.person.get_unknown_person),\n related_name=\"faces\",\n to=\"api.Photo\",\n ),\n ),\n migrations.AddField(\n model_name=\"albumuser\",\n name=\"cover_photos\",\n field=models.ManyToManyField(\n related_name=\"album_user_cover_photos\", to=\"api.Photo\"\n ),\n ),\n migrations.AddField(\n model_name=\"albumuser\",\n name=\"owner\",\n field=models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n migrations.AddField(\n model_name=\"albumuser\",\n name=\"photos\",\n field=models.ManyToManyField(to=\"api.Photo\"),\n ),\n migrations.AddField(\n model_name=\"albumuser\",\n name=\"shared_to\",\n field=models.ManyToManyField(\n related_name=\"album_user_shared_to\", to=settings.AUTH_USER_MODEL\n ),\n ),\n migrations.AddField(\n model_name=\"albumthing\",\n name=\"cover_photos\",\n field=models.ManyToManyField(\n related_name=\"album_thing_cover_photos\", to=\"api.Photo\"\n ),\n ),\n migrations.AddField(\n model_name=\"albumthing\",\n name=\"owner\",\n field=models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n migrations.AddField(\n model_name=\"albumthing\",\n name=\"photos\",\n field=models.ManyToManyField(to=\"api.Photo\"),\n ),\n migrations.AddField(\n model_name=\"albumthing\",\n name=\"shared_to\",\n field=models.ManyToManyField(\n related_name=\"album_thing_shared_to\", to=settings.AUTH_USER_MODEL\n ),\n ),\n migrations.AddField(\n model_name=\"albumplace\",\n name=\"cover_photos\",\n field=models.ManyToManyField(\n related_name=\"album_place_cover_photos\", to=\"api.Photo\"\n ),\n ),\n migrations.AddField(\n model_name=\"albumplace\",\n name=\"owner\",\n field=models.ForeignKey(\n default=None,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n migrations.AddField(\n model_name=\"albumplace\",\n name=\"photos\",\n field=models.ManyToManyField(to=\"api.Photo\"),\n ),\n migrations.AddField(\n model_name=\"albumplace\",\n name=\"shared_to\",\n field=models.ManyToManyField(\n related_name=\"album_place_shared_to\", to=settings.AUTH_USER_MODEL\n ),\n ),\n migrations.AddField(\n model_name=\"albumdate\",\n name=\"photos\",\n field=models.ManyToManyField(to=\"api.Photo\"),\n ),\n migrations.AddField(\n model_name=\"albumdate\",\n name=\"shared_to\",\n field=models.ManyToManyField(\n related_name=\"album_date_shared_to\", to=settings.AUTH_USER_MODEL\n ),\n ),\n migrations.AddField(\n model_name=\"albumauto\",\n name=\"photos\",\n field=models.ManyToManyField(to=\"api.Photo\"),\n ),\n migrations.AddField(\n model_name=\"albumauto\",\n name=\"shared_to\",\n field=models.ManyToManyField(\n related_name=\"album_auto_shared_to\", to=settings.AUTH_USER_MODEL\n ),\n ),\n migrations.AlterUniqueTogether(\n name=\"albumuser\",\n unique_together={(\"title\", \"owner\")},\n ),\n migrations.AlterUniqueTogether(\n name=\"albumthing\",\n unique_together={(\"title\", \"owner\")},\n ),\n migrations.AlterUniqueTogether(\n name=\"albumplace\",\n unique_together={(\"title\", \"owner\")},\n ),\n migrations.AlterUniqueTogether(\n name=\"albumdate\",\n unique_together={(\"date\", \"owner\")},\n ),\n migrations.AlterUniqueTogether(\n name=\"albumauto\",\n unique_together={(\"timestamp\", \"owner\")},\n ),\n migrations.RemoveField(model_name=\"albumplace\", name=\"cover_photos\"),\n migrations.RemoveField(model_name=\"albumthing\", name=\"cover_photos\"),\n migrations.RemoveField(model_name=\"albumuser\", name=\"cover_photos\"),\n ]\n" }, { "path": "api/migrations/0002_add_confidence.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0001_initial\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"User\",\n name=\"confidence\",\n field=models.FloatField(default=0.1, db_index=True),\n )\n ]\n" }, { "path": "api/migrations/0003_remove_unused_thumbs.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0002_add_confidence\"),\n ]\n\n operations = [\n migrations.RemoveField(model_name=\"Photo\", name=\"thumbnail_tiny\"),\n migrations.RemoveField(model_name=\"Photo\", name=\"thumbnail_small\"),\n migrations.RemoveField(model_name=\"Photo\", name=\"thumbnail\"),\n migrations.RemoveField(model_name=\"Photo\", name=\"square_thumbnail_tiny\"),\n migrations.RemoveField(model_name=\"Photo\", name=\"square_thumbnail_big\"),\n ]\n" }, { "path": "api/migrations/0004_fix_album_thing_constraint.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0003_remove_unused_thumbs\"),\n ]\n\n operations = [\n migrations.AlterUniqueTogether(\n name=\"albumthing\",\n unique_together=set([]),\n ),\n migrations.AddConstraint(\n model_name=\"albumthing\",\n constraint=models.UniqueConstraint(\n fields=[\"title\", \"thing_type\", \"owner\"],\n name=\"unique AlbumThing\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0005_add_video_to_photo.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0004_fix_album_thing_constraint\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"Photo\", name=\"video\", field=models.BooleanField(default=False)\n )\n ]\n" }, { "path": "api/migrations/0006_migrate_to_boolean_field.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0005_add_video_to_photo\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"Face\",\n name=\"person_label_is_inferred\",\n field=models.BooleanField(null=True, db_index=True),\n )\n ]\n" }, { "path": "api/migrations/0007_migrate_to_json_field.py", "content": "import json\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0006_migrate_to_boolean_field\"),\n ]\n\n def forwards_func(apps, schema):\n Photo = apps.get_model(\"api\", \"Photo\")\n for obj in Photo.objects.all():\n try:\n obj.image_paths.append(obj.image_path)\n obj.save()\n except json.decoder.JSONDecodeError:\n print(\"Cannot convert {} object\".format(obj.image_path))\n\n operations = [\n migrations.AddField(\n model_name=\"Photo\",\n name=\"image_paths\",\n field=models.JSONField(db_index=True, default=list),\n ),\n migrations.RunPython(forwards_func),\n ]\n" }, { "path": "api/migrations/0008_remove_image_path.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0007_migrate_to_json_field\"),\n ]\n\n operations = [\n migrations.RemoveField(model_name=\"Photo\", name=\"image_path\"),\n ]\n" }, { "path": "api/migrations/0009_add_aspect_ratio.py", "content": "import exiftool\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0008_remove_image_path\"),\n ]\n\n def forwards_func(apps, schema):\n Photo = apps.get_model(\"api\", \"Photo\")\n with exiftool.ExifTool() as et:\n for obj in Photo.objects.all():\n if obj.thumbnail_big:\n try:\n height = et.get_tag(\"ImageHeight\", obj.thumbnail_big.path)\n width = et.get_tag(\"ImageWidth\", obj.thumbnail_big.path)\n obj.aspect_ratio = round((width / height), 2)\n obj.save()\n except Exception:\n print(\"Cannot convert {} object\".format(obj))\n\n operations = [\n migrations.AddField(\n model_name=\"Photo\",\n name=\"aspect_ratio\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.RunPython(forwards_func),\n ]\n" }, { "path": "api/migrations/0009_add_clip_embedding_field.py", "content": "from django.contrib.postgres.fields import ArrayField\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0008_remove_image_path\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"Photo\",\n name=\"clip_embeddings\",\n field=ArrayField(\n models.FloatField(blank=True, null=True), size=512, null=True\n ),\n ),\n migrations.AddField(\n model_name=\"Photo\",\n name=\"clip_embeddings_magnitude\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"User\",\n name=\"semantic_search_topk\",\n field=models.IntegerField(default=0, null=False),\n ),\n migrations.RemoveField(\n model_name=\"Photo\",\n name=\"encoding\",\n ),\n ]\n" }, { "path": "api/migrations/0010_merge_20210725_1547.py", "content": "# Generated by Django 3.1.8 on 2021-07-25 21:47\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0009_add_aspect_ratio\"),\n (\"api\", \"0009_add_clip_embedding_field\"),\n ]\n\n operations = []\n" }, { "path": "api/migrations/0011_a_add_rating.py", "content": "# Generated by Django 3.1.8 on 2021-08-06 11:32\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0010_merge_20210725_1547\"),\n ]\n\n run_before = [(\"api\", \"0011_b_migrate_favorited_to_rating\")]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"rating\",\n field=models.IntegerField(db_index=True, default=0),\n ),\n ]\n" }, { "path": "api/migrations/0011_b_migrate_favorited_to_rating.py", "content": "# Generated by Django 3.1.8 on 2021-08-06 11:32\n\nfrom django.db import migrations\n\n\ndef favorited_to_rating(apps, schema_editor):\n Photo = apps.get_model(\"api\", \"Photo\")\n for photo in Photo.objects.all():\n photo.rating = 4 if photo.favorited else 0\n photo.save()\n\n\ndef rating_to_favorited(apps, schema_editor):\n Photo = apps.get_model(\"api\", \"Photo\")\n for photo in Photo.objects.all():\n photo.favorited = photo.rating >= 4\n photo.save()\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0011_a_add_rating\"),\n ]\n\n run_before = [(\"api\", \"0011_c_remove_favorited\")]\n\n operations = [migrations.RunPython(favorited_to_rating, rating_to_favorited)]\n" }, { "path": "api/migrations/0011_c_remove_favorited.py", "content": "# Generated by Django 3.1.8 on 2021-08-06 11:32\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0011_b_migrate_favorited_to_rating\"),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"favorited\",\n ),\n ]\n" }, { "path": "api/migrations/0012_add_favorite_min_rating.py", "content": "# Generated by Django 3.1.8 on 2021-08-08 17:05\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0011_c_remove_favorited\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"favorite_min_rating\",\n field=models.IntegerField(db_index=True, default=4),\n ),\n ]\n" }, { "path": "api/migrations/0013_add_image_scale_and_misc.py", "content": "# Generated by Django 3.1.8 on 2021-09-09 16:06\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0012_add_favorite_min_rating\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"image_scale\",\n field=models.FloatField(default=1),\n ),\n migrations.AlterField(\n model_name=\"albumdate\",\n name=\"location\",\n field=models.JSONField(blank=True, db_index=True, null=True),\n ),\n migrations.AlterField(\n model_name=\"face\",\n name=\"image\",\n field=models.ImageField(null=True, upload_to=\"faces\"),\n ),\n migrations.AlterField(\n model_name=\"face\",\n name=\"photo\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.CASCADE,\n related_name=\"faces\",\n to=\"api.photo\",\n ),\n ),\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n ]\n ),\n ),\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"result\",\n field=models.JSONField(default={\"progress\": {\"target\": 0, \"current\": 0}}),\n ),\n migrations.AlterField(\n model_name=\"photo\",\n name=\"captions_json\",\n field=models.JSONField(blank=True, db_index=True, null=True),\n ),\n migrations.AlterField(\n model_name=\"photo\",\n name=\"exif_json\",\n field=models.JSONField(blank=True, null=True),\n ),\n migrations.AlterField(\n model_name=\"photo\",\n name=\"geolocation_json\",\n field=models.JSONField(blank=True, db_index=True, null=True),\n ),\n migrations.AlterField(\n model_name=\"photo\",\n name=\"image_paths\",\n field=models.JSONField(default=list),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"first_name\",\n field=models.CharField(\n blank=True, max_length=150, verbose_name=\"first name\"\n ),\n ),\n ]\n" }, { "path": "api/migrations/0014_add_save_metadata_to_disk.py", "content": "# Generated by Django 3.1.8 on 2021-08-08 17:05\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0013_add_image_scale_and_misc\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"save_metadata_to_disk\",\n field=models.TextField(\n choices=[\n (\"OFF\", \"Off\"),\n (\"MEDIA_FILE\", \"Media File\"),\n (\"SIDECAR_FILE\", \"Sidecar File\"),\n ],\n default=\"OFF\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0015_add_dominant_color.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0014_add_save_metadata_to_disk\"),\n ]\n operations = [\n migrations.AddField(\n model_name=\"Photo\",\n name=\"dominant_color\",\n field=models.TextField(blank=True, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0016_add_transcode_videos.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0015_add_dominant_color\"),\n ]\n operations = [\n migrations.AddField(\n model_name=\"User\",\n name=\"transcode_videos\",\n field=models.BooleanField(default=False),\n ),\n ]\n" }, { "path": "api/migrations/0017_add_cover_photo.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0016_add_transcode_videos\"),\n ]\n operations = [\n migrations.AddField(\n model_name=\"Person\",\n name=\"cover_photo\",\n field=models.ForeignKey(\n to=\"api.Photo\",\n related_name=\"person\",\n on_delete=models.PROTECT,\n blank=False,\n null=True,\n ),\n ),\n ]\n" }, { "path": "api/migrations/0018_user_config_datetime_rules.py", "content": "# Generated by Django 3.1.14 on 2022-01-24 17:11\n\nfrom django.db import migrations, models\n\nimport api.models.user\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0017_add_cover_photo\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"config_datetime_rules\",\n field=models.JSONField(\n default=api.models.user.get_default_config_datetime_rules\n ),\n ),\n ]\n" }, { "path": "api/migrations/0019_change_config_datetime_rules.py", "content": "# Generated by Django 3.1.14 on 2022-01-24 17:11\n\nfrom django.db import migrations, models\n\nimport api.models.user\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0018_user_config_datetime_rules\"),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name=\"user\",\n name=\"config_datetime_rules\",\n ),\n migrations.AddField(\n model_name=\"user\",\n name=\"datetime_rules\",\n field=models.JSONField(\n default=api.models.user.get_default_config_datetime_rules\n ),\n ),\n ]\n" }, { "path": "api/migrations/0020_add_default_timezone.py", "content": "# Generated by Django 3.1.14 on 2022-01-24 17:11\n\nimport pytz\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0019_change_config_datetime_rules\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"default_timezone\",\n field=models.TextField(\n choices=[(x, x) for x in pytz.all_timezones],\n default=\"UTC\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0021_remove_photo_image.py", "content": "# Generated by Django 3.1.14 on 2022-02-01 22:42\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0020_add_default_timezone\"),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"image\",\n ),\n ]\n" }, { "path": "api/migrations/0022_photo_video_length.py", "content": "# Generated by Django 3.1.14 on 2022-02-20 11:16\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0021_remove_photo_image\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"video_length\",\n field=models.TextField(blank=True, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0023_photo_deleted.py", "content": "# Generated by Django 3.1.14 on 2022-02-23 21:29\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0022_photo_video_length\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"deleted\",\n field=models.BooleanField(db_index=True, default=False),\n ),\n ]\n" }, { "path": "api/migrations/0024_photo_timestamp.py", "content": "# Generated by Django 3.1.14 on 2022-03-18 10:35\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0023_photo_deleted\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"timestamp\",\n field=models.DateTimeField(blank=True, db_index=True, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0025_add_cover_photo.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0024_photo_timestamp\"),\n ]\n operations = [\n migrations.AddField(\n model_name=\"AlbumUser\",\n name=\"cover_photo\",\n field=models.ForeignKey(\n to=\"api.Photo\",\n related_name=\"album_user\",\n on_delete=models.PROTECT,\n blank=False,\n null=True,\n ),\n ),\n ]\n" }, { "path": "api/migrations/0026_add_cluster_info.py", "content": "# Generated by Django 3.1.14 on 2022-07-15 21:19\n\nimport django.db.models.deletion\nimport django.db.models.manager\nfrom django.conf import settings\nfrom django.db import migrations, models\n\nimport api.models.person\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0025_add_cover_photo\"),\n ]\n\n operations = [\n migrations.AlterModelManagers(\n name=\"albumdate\",\n managers=[\n (\"visible\", django.db.models.manager.Manager()),\n ],\n ),\n migrations.RemoveField(\n model_name=\"person\",\n name=\"cluster_id\",\n ),\n migrations.AlterField(\n model_name=\"person\",\n name=\"mean_face_encoding\",\n field=models.TextField(default=\"default\"),\n ),\n migrations.RemoveField(\n model_name=\"person\",\n name=\"mean_face_encoding\",\n ),\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n (8, \"Find Similar Faces\"),\n ]\n ),\n ),\n migrations.CreateModel(\n name=\"Cluster\",\n fields=[\n (\n \"id\",\n models.AutoField(\n auto_created=True,\n primary_key=True,\n serialize=False,\n verbose_name=\"ID\",\n ),\n ),\n (\"mean_face_encoding\", models.TextField()),\n (\"cluster_id\", models.IntegerField(null=True)),\n (\"name\", models.TextField(null=True)),\n (\n \"person\",\n models.ForeignKey(\n blank=True,\n null=True,\n on_delete=models.SET(api.models.person.get_unknown_person),\n related_name=\"clusters\",\n to=\"api.person\",\n ),\n ),\n (\n \"owner\",\n models.ForeignKey(\n default=None,\n null=True,\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ],\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"cluster\",\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=models.SET(api.models.cluster.get_unknown_cluster),\n related_name=\"faces\",\n to=\"api.cluster\",\n ),\n ),\n migrations.AddField(\n model_name=\"person\",\n name=\"cluster_owner\",\n field=models.ForeignKey(\n default=None,\n null=True,\n related_name=\"owner\",\n on_delete=models.SET(api.models.user.get_deleted_user),\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ]\n" }, { "path": "api/migrations/0027_rename_unknown_person.py", "content": "# Generated by Django 3.1.14 on 2022-07-17 19:07\nfrom django.db import migrations\n\nUNKNOWN_PERSON_NAME = \"Unknown - Other\"\nKIND_UNKNOWN = \"UNKNOWN\"\n\n\ndef migrate_unknown(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n person: Person\n try:\n person = Person.objects.get(name=\"unknown\")\n person.name = UNKNOWN_PERSON_NAME\n person.kind = KIND_UNKNOWN\n person.save()\n except Person.DoesNotExist:\n unknown_person: Person = Person.objects.get_or_create(\n name=UNKNOWN_PERSON_NAME, cluster_owner=None, kind=KIND_UNKNOWN\n )[0]\n if unknown_person.kind != KIND_UNKNOWN:\n unknown_person.kind = KIND_UNKNOWN\n unknown_person.save()\n\n\ndef unmigrate_unknown(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n try:\n person: Person = Person.objects.get(name=UNKNOWN_PERSON_NAME)\n person.name = \"unknown\"\n person.kind = \"\"\n person.save()\n except Person.DoesNotExist:\n pass\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0026_add_cluster_info\"),\n ]\n\n operations = [migrations.RunPython(migrate_unknown, unmigrate_unknown)]\n" }, { "path": "api/migrations/0028_add_metadata_fields.py", "content": "# Generated by Django 3.1.14 on 2022-07-30 11:47\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0027_rename_unknown_person\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"camera\",\n field=models.TextField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"digitalZoomRatio\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"focalLength35Equivalent\",\n field=models.IntegerField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"focal_length\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"fstop\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"height\",\n field=models.IntegerField(default=0),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"iso\",\n field=models.IntegerField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"lens\",\n field=models.TextField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"shutter_speed\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"size\",\n field=models.IntegerField(default=0),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"subjectDistance\",\n field=models.FloatField(blank=True, null=True),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"width\",\n field=models.IntegerField(default=0),\n ),\n ]\n" }, { "path": "api/migrations/0029_change_to_text_field.py", "content": "# Generated by Django 3.1.14 on 2022-07-31 11:35\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0028_add_metadata_fields\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"photo\",\n name=\"shutter_speed\",\n field=models.TextField(blank=True, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0030_user_confidence_person.py", "content": "# Generated by Django 3.1.14 on 2022-08-08 13:39\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0029_change_to_text_field\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"confidence_person\",\n field=models.FloatField(default=0.9),\n ),\n ]\n" }, { "path": "api/migrations/0031_remove_account.py", "content": "# Generated by Django 3.1.14 on 2022-09-01 16:28\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0030_user_confidence_person\"),\n ]\n\n operations = [\n migrations.AlterModelManagers(\n name=\"albumdate\",\n managers=[],\n ),\n migrations.RemoveField(\n model_name=\"person\",\n name=\"account\",\n ),\n ]\n" }, { "path": "api/migrations/0032_always_have_owner.py", "content": "# Generated by Django 3.1.8 on 2021-08-06 11:32\n\nfrom django.db import migrations\n\n\ndef add_cluster_owner(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n for person in Person.objects.all():\n if person.faces.first():\n person.cluster_owner = person.faces.first().photo.owner\n person.save()\n\n\ndef remove_cluster_owner(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n for person in Person.objects.all():\n person.cluster_owner = None\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0031_remove_account\"),\n ]\n\n operations = [migrations.RunPython(add_cluster_owner, remove_cluster_owner)]\n" }, { "path": "api/migrations/0033_add_post_delete_person.py", "content": "# Generated by Django 3.1.14 on 2022-09-02 10:23\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0032_always_have_owner\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"face\",\n name=\"person\",\n field=models.ForeignKey(\n on_delete=django.db.models.deletion.DO_NOTHING,\n related_name=\"faces\",\n to=\"api.person\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0034_allow_deleting_person.py", "content": "# Generated by Django 3.1.14 on 2022-09-02 10:26\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0033_add_post_delete_person\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"person\",\n name=\"cover_photo\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"person\",\n to=\"api.photo\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0035_add_files_model.py", "content": "# Generated by Django 3.1.14 on 2022-11-09 17:35\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0034_allow_deleting_person\"),\n ]\n\n operations = [\n migrations.CreateModel(\n name=\"File\",\n fields=[\n (\n \"hash\",\n models.CharField(max_length=64, primary_key=True, serialize=False),\n ),\n (\"path\", models.TextField(blank=True, null=True)),\n (\n \"type\",\n models.PositiveIntegerField(\n blank=True,\n choices=[\n (1, \"Image\"),\n (2, \"Video\"),\n (3, \"Metadata File e.g. XMP\"),\n (4, \"Raw File\"),\n ],\n ),\n ),\n ],\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"files\",\n field=models.ManyToManyField(to=\"api.File\"),\n ),\n migrations.AddField(\n model_name=\"photo\",\n name=\"main_file\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.PROTECT,\n related_name=\"main_photo\",\n to=\"api.file\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0036_handle_missing_files.py", "content": "# Generated by Django 3.1.14 on 2022-11-10 08:41\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0035_add_files_model\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"file\",\n name=\"missing\",\n field=models.BooleanField(default=False),\n ),\n migrations.AlterField(\n model_name=\"file\",\n name=\"type\",\n field=models.PositiveIntegerField(\n blank=True,\n choices=[\n (1, \"Image\"),\n (2, \"Video\"),\n (3, \"Metadata File e.g. XMP\"),\n (4, \"Raw File\"),\n (5, \"Unknown\"),\n ],\n ),\n ),\n ]\n" }, { "path": "api/migrations/0037_migrate_to_files.py", "content": "import os\n\nfrom django.db import migrations\n\nfrom api.models.file import is_metadata, is_raw, is_video\n\nIMAGE = 1\nVIDEO = 2\nMETADATA_FILE = 3\nRAW_FILE = 4\nUNKNOWN = 5\n\n\ndef find_out_type(path):\n if is_raw(path):\n return RAW_FILE\n if is_video(path):\n return VIDEO\n if is_metadata(path):\n return METADATA_FILE\n return IMAGE\n\n\ndef migrate_to_files(apps, schema_editor):\n Photo = apps.get_model(\"api\", \"Photo\")\n File = apps.get_model(\"api\", \"File\")\n for photo in Photo.objects.all():\n if photo.image_paths:\n for path in photo.image_paths:\n file: File = File()\n file.path = path\n if os.path.exists(path):\n file.type = find_out_type(path)\n else:\n file.type = UNKNOWN\n if photo.video:\n file.type = VIDEO\n file.missing = True\n # This is fine, because at this point all files that belong to a photo have the same hash\n file.hash = photo.image_hash\n file.save()\n photo.files.add(file)\n photo.save()\n # handle missing photos\n else:\n file: File = File()\n file.path = None\n file.type = UNKNOWN\n file.missing = True\n file.hash = photo.image_hash\n file.save()\n photo.files.add(file)\n photo.save()\n\n\ndef remove_files(apps, schema_editor):\n File = apps.get_model(\"api\", \"File\")\n for file in File.objects.all():\n file.delete()\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0036_handle_missing_files\"),\n ]\n\n operations = [migrations.RunPython(migrate_to_files, remove_files)]\n" }, { "path": "api/migrations/0038_add_main_file.py", "content": "from django.db import migrations\n\nfrom api.models.file import is_metadata, is_raw, is_video\n\nIMAGE = 1\nVIDEO = 2\nMETADATA_FILE = 3\nRAW_FILE = 4\nUNKNOWN = 5\n\n\ndef find_out_type(path):\n if is_raw(path):\n return RAW_FILE\n if is_video(path):\n return VIDEO\n if is_metadata(path):\n return METADATA_FILE\n return IMAGE\n\n\ndef add_main_file(apps, schema_editor):\n Photo = apps.get_model(\"api\", \"Photo\")\n for photo in Photo.objects.all():\n if photo.files.count() > 0:\n photo.main_file = photo.files.first()\n photo.save()\n\n\ndef remove_main_file(apps, schema_editor):\n Photo = apps.get_model(\"api\", \"Photo\")\n for photo in Photo.objects.all():\n photo.main_file = None\n photo.save()\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0037_migrate_to_files\"),\n ]\n\n operations = [migrations.RunPython(add_main_file, remove_main_file)]\n" }, { "path": "api/migrations/0039_remove_photo_image_paths.py", "content": "# Generated by Django 3.1.14 on 2022-12-21 09:24\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0038_add_main_file\"),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"image_paths\",\n ),\n ]\n" }, { "path": "api/migrations/0040_add_user_public_sharing_flag.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0039_remove_photo_image_paths\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"public_sharing\",\n field=models.BooleanField(default=False),\n ),\n ]\n" }, { "path": "api/migrations/0041_apply_user_enum_for_person.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n def apply_enum(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n for person in Person.objects.filter(kind=\"\").all():\n person.kind = \"USER\"\n person.save()\n\n def remove_enum(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n for person in Person.objects.filter(kind=\"\").all():\n person.kind = \"\"\n person.save()\n\n dependencies = [\n (\"api\", \"0040_add_user_public_sharing_flag\"),\n ]\n\n operations = [migrations.RunPython(apply_enum, remove_enum)]\n" }, { "path": "api/migrations/0042_alter_albumuser_cover_photo_alter_photo_main_file.py", "content": "# Generated by Django 4.2rc1 on 2023-04-04 09:14\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0041_apply_user_enum_for_person\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"albumuser\",\n name=\"cover_photo\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"album_user\",\n to=\"api.photo\",\n ),\n ),\n migrations.AlterField(\n model_name=\"photo\",\n name=\"main_file\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"main_photo\",\n to=\"api.file\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0043_alter_photo_size.py", "content": "# Generated by Django 4.2rc1 on 2023-04-05 07:34\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0042_alter_albumuser_cover_photo_alter_photo_main_file\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"photo\",\n name=\"size\",\n field=models.BigIntegerField(default=0),\n ),\n ]\n" }, { "path": "api/migrations/0044_alter_cluster_person_alter_person_cluster_owner.py", "content": "# Generated by Django 4.2rc1 on 2023-04-07 19:02\n\nimport django.db.models.deletion\nfrom django.conf import settings\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0043_alter_photo_size\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"cluster\",\n name=\"person\",\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"clusters\",\n to=\"api.person\",\n ),\n ),\n migrations.AlterField(\n model_name=\"person\",\n name=\"cluster_owner\",\n field=models.ForeignKey(\n default=None,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"owner\",\n to=settings.AUTH_USER_MODEL,\n ),\n ),\n ]\n" }, { "path": "api/migrations/0045_alter_face_cluster.py", "content": "# Generated by Django 4.2rc1 on 2023-04-07 19:15\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0044_alter_cluster_person_alter_person_cluster_owner\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"face\",\n name=\"cluster\",\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"faces\",\n to=\"api.cluster\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0046_add_embedded_media.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0045_alter_face_cluster\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"file\",\n name=\"embedded_media\",\n field=models.ManyToManyField(\"File\"),\n ),\n ]\n" }, { "path": "api/migrations/0047_alter_file_embedded_media.py", "content": "# Generated by Django 4.2rc1 on 2023-04-15 10:42\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0046_add_embedded_media\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"file\",\n name=\"embedded_media\",\n field=models.ManyToManyField(to=\"api.file\"),\n ),\n ]\n" }, { "path": "api/migrations/0048_fix_null_height.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0047_alter_file_embedded_media\"),\n ]\n\n operations = [\n migrations.RunSQL(\"UPDATE api_photo SET height=0 WHERE height IS NULL;\")\n ]\n" }, { "path": "api/migrations/0049_fix_metadata_files_as_main_files.py", "content": "from django.db import migrations\n\n\ndef delete_photos_with_metadata_as_main(apps, schema_editor):\n Photo = apps.get_model(\"api\", \"Photo\")\n for photo in Photo.objects.filter(main_file__type=3):\n photo.delete()\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0048_fix_null_height\"),\n ]\n\n operations = [\n migrations.RunPython(delete_photos_with_metadata_as_main),\n ]\n" }, { "path": "api/migrations/0050_person_face_count.py", "content": "# Generated by Django 4.2.1 on 2023-06-20 09:46\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0049_fix_metadata_files_as_main_files\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"person\",\n name=\"face_count\",\n field=models.IntegerField(default=0),\n ),\n ]\n" }, { "path": "api/migrations/0051_set_person_defaults.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n def apply_default(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n User = apps.get_model(\"api\", \"User\")\n\n for person in Person.objects.filter(kind=\"USER\").all():\n number_of_faces = person.faces.filter(\n photo__hidden=False,\n photo__deleted=False,\n photo__owner=person.cluster_owner.id,\n ).count()\n if not person.cover_photo and number_of_faces > 0:\n person.cover_photo = (\n person.faces.filter(\n photo__hidden=False,\n photo__deleted=False,\n photo__owner=person.cluster_owner.id,\n )\n .first()\n .photo\n )\n confidence_person = (\n User.objects.filter(id=person.cluster_owner.id)\n .first()\n .confidence_person\n )\n person.face_count = person.faces.filter(\n photo__hidden=False,\n photo__deleted=False,\n photo__owner=person.cluster_owner.id,\n person_label_probability__gte=confidence_person,\n ).count()\n person.save()\n\n def remove_default(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n for person in Person.objects.all():\n person.face_count = 0\n person.save()\n\n dependencies = [\n (\"api\", \"0050_person_face_count\"),\n ]\n\n operations = [migrations.RunPython(apply_default, remove_default)]\n" }, { "path": "api/migrations/0052_alter_person_name.py", "content": "# Generated by Django 4.2.1 on 2023-06-26 12:14\n\nimport django.core.validators\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0051_set_person_defaults\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"person\",\n name=\"name\",\n field=models.CharField(\n max_length=128,\n validators=[django.core.validators.MinLengthValidator(1)],\n ),\n ),\n ]\n" }, { "path": "api/migrations/0053_user_confidence_unknown_face_and_more.py", "content": "# Generated by Django 4.2.1 on 2023-07-09 11:08\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0052_alter_person_name\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"confidence_unknown_face\",\n field=models.FloatField(default=0.5),\n ),\n migrations.AddField(\n model_name=\"user\",\n name=\"face_recognition_model\",\n field=models.TextField(\n choices=[(\"HOG\", \"Hog\"), (\"CNN\", \"Cnn\")], default=\"HOG\"\n ),\n ),\n migrations.AddField(\n model_name=\"user\",\n name=\"min_cluster_size\",\n field=models.IntegerField(default=0),\n ),\n ]\n" }, { "path": "api/migrations/0054_user_cluster_selection_epsilon_user_min_samples.py", "content": "# Generated by Django 4.2.1 on 2023-07-11 11:06\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0053_user_confidence_unknown_face_and_more\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"cluster_selection_epsilon\",\n field=models.FloatField(default=0.05),\n ),\n migrations.AddField(\n model_name=\"user\",\n name=\"min_samples\",\n field=models.IntegerField(default=1),\n ),\n ]\n" }, { "path": "api/migrations/0055_alter_longrunningjob_job_type.py", "content": "# Generated by Django 4.2.6 on 2023-10-27 13:01\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0054_user_cluster_selection_epsilon_user_min_samples\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n (8, \"Find Similar Faces\"),\n (9, \"Download Selected Photos\"),\n ]\n ),\n ),\n ]\n" }, { "path": "api/migrations/0056_user_llm_settings_alter_longrunningjob_job_type.py", "content": "# Generated by Django 4.2.8 on 2023-12-21 11:16\n\nfrom django.db import migrations, models\n\nimport api.models.user\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0055_alter_longrunningjob_job_type\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"llm_settings\",\n field=models.JSONField(default=api.models.user.get_default_llm_settings),\n ),\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n (8, \"Find Similar Faces\"),\n (9, \"Download Selected Photos\"),\n (10, \"Download Models\"),\n ]\n ),\n ),\n ]\n" }, { "path": "api/migrations/0057_remove_face_image_path_and_more.py", "content": "# Generated by Django 4.2.8 on 2024-01-10 17:12\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0056_user_llm_settings_alter_longrunningjob_job_type\"),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name=\"face\",\n name=\"image_path\",\n ),\n migrations.AlterField(\n model_name=\"face\",\n name=\"person_label_is_inferred\",\n field=models.BooleanField(db_index=True, default=False),\n ),\n ]\n" }, { "path": "api/migrations/0058_alter_user_avatar_alter_user_nextcloud_app_password_and_more.py", "content": "# Generated by Django 4.2.9 on 2024-02-02 16:36\n\nimport django_cryptography.fields\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0057_remove_face_image_path_and_more\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"user\",\n name=\"avatar\",\n field=models.ImageField(blank=True, null=True, upload_to=\"avatars\"),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_app_password\",\n field=django_cryptography.fields.encrypt(\n models.CharField(blank=True, default=None, max_length=64, null=True)\n ),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_scan_directory\",\n field=models.CharField(\n blank=True, db_index=True, max_length=512, null=True\n ),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_server_address\",\n field=models.CharField(blank=True, default=None, max_length=200, null=True),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_username\",\n field=models.CharField(blank=True, default=None, max_length=64, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0059_person_cover_face.py", "content": "# Generated by Django 4.2.11 on 2024-03-29 16:03\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0058_alter_user_avatar_alter_user_nextcloud_app_password_and_more\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"person\",\n name=\"cover_face\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"face\",\n to=\"api.face\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0060_apply_default_face_cover.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n def apply_default(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n\n for person in Person.objects.filter(kind=\"USER\").all():\n if not person.cover_face and person.faces.count() > 0:\n person.cover_face = person.faces.first()\n person.save()\n if (\n not person.cover_face\n and person.cover_photo\n and person.cover_photo.faces.count() > 0\n ):\n person.cover_face = person.cover_photo.faces.filter(\n person__name=person.name\n ).first()\n person.save()\n\n def remove_default(apps, schema_editor):\n Person = apps.get_model(\"api\", \"Person\")\n for person in Person.objects.all():\n person.cover_face = None\n\n dependencies = [\n (\"api\", \"0059_person_cover_face\"),\n ]\n\n operations = [migrations.RunPython(apply_default, remove_default)]\n" }, { "path": "api/migrations/0061_alter_person_name.py", "content": "# Generated by Django 4.2.11 on 2024-03-29 17:20\n\nimport django.core.validators\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0060_apply_default_face_cover\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"person\",\n name=\"name\",\n field=models.CharField(\n db_index=True,\n max_length=128,\n validators=[django.core.validators.MinLengthValidator(1)],\n ),\n ),\n ]\n" }, { "path": "api/migrations/0062_albumthing_cover_photos.py", "content": "# Generated by Django 4.2.11 on 2024-03-29 17:33\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0061_alter_person_name\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"albumthing\",\n name=\"cover_photos\",\n field=models.ManyToManyField(\n related_name=\"album_thing_cover_photos\", to=\"api.photo\"\n ),\n ),\n ]\n" }, { "path": "api/migrations/0063_apply_default_album_things_cover.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n def apply_default(apps, schema_editor):\n AlbumThing = apps.get_model(\"api\", \"AlbumThing\")\n\n for thing in AlbumThing.objects.all():\n if thing.photos.count() > 0:\n thing.cover_photos.add(*thing.photos.all()[:4])\n thing.save()\n\n def remove_default(apps, schema_editor):\n AlbumThing = apps.get_model(\"api\", \"AlbumThing\")\n for thing in AlbumThing.objects.all():\n thing.cover_photos = None\n thing.save()\n\n dependencies = [\n (\"api\", \"0062_albumthing_cover_photos\"),\n ]\n\n operations = [migrations.RunPython(apply_default, remove_default)]\n" }, { "path": "api/migrations/0064_albumthing_photo_count.py", "content": "# Generated by Django 4.2.11 on 2024-03-30 13:20\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0063_apply_default_album_things_cover\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"albumthing\",\n name=\"photo_count\",\n field=models.IntegerField(default=0),\n ),\n ]\n" }, { "path": "api/migrations/0065_apply_default_photo_count.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n def apply_default(apps, schema_editor):\n AlbumThing = apps.get_model(\"api\", \"AlbumThing\")\n\n for thing in AlbumThing.objects.all():\n thing.photo_count = thing.photos.filter(hidden=False).count()\n thing.save()\n\n def remove_default(apps, schema_editor):\n AlbumThing = apps.get_model(\"api\", \"AlbumThing\")\n for thing in AlbumThing.objects.all():\n thing.photo_count = 0\n thing.save()\n\n dependencies = [\n (\"api\", \"0064_albumthing_photo_count\"),\n ]\n\n operations = [migrations.RunPython(apply_default, remove_default)]\n" }, { "path": "api/migrations/0066_photo_last_modified_alter_longrunningjob_job_type.py", "content": "# Generated by Django 4.2.13 on 2024-06-12 15:09\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0065_apply_default_photo_count\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"last_modified\",\n field=models.DateTimeField(auto_now=True),\n ),\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n (8, \"Find Similar Faces\"),\n (9, \"Download Selected Photos\"),\n (10, \"Download Models\"),\n (11, \"Add Geolocation\"),\n (12, \"Generate Tags\"),\n ]\n ),\n ),\n ]\n" }, { "path": "api/migrations/0067_alter_longrunningjob_job_type.py", "content": "# Generated by Django 4.2.13 on 2024-06-16 15:40\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0066_photo_last_modified_alter_longrunningjob_job_type\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n (8, \"Find Similar Faces\"),\n (9, \"Download Selected Photos\"),\n (10, \"Download Models\"),\n (11, \"Add Geolocation\"),\n (12, \"Generate Tags\"),\n (13, \"Generate Face Embeddings\"),\n ]\n ),\n ),\n ]\n" }, { "path": "api/migrations/0068_remove_longrunningjob_result_and_more.py", "content": "# Generated by Django 4.2.13 on 2024-06-18 13:18\nfrom django.db import migrations, models\n\n\ndef copy_progress_data(apps, schema_editor):\n LongRunningJob = apps.get_model(\"api\", \"LongRunningJob\")\n for job in LongRunningJob.objects.all():\n result = job.result\n job.progress_current = result[\"progress\"][\"current\"]\n job.progress_target = result[\"progress\"][\"target\"]\n job.save()\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0067_alter_longrunningjob_job_type\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"longrunningjob\",\n name=\"progress_current\",\n field=models.PositiveIntegerField(default=0),\n ),\n migrations.AddField(\n model_name=\"longrunningjob\",\n name=\"progress_target\",\n field=models.PositiveIntegerField(default=0),\n ),\n migrations.RunPython(copy_progress_data),\n migrations.RemoveField(\n model_name=\"longrunningjob\",\n name=\"result\",\n ),\n ]\n" }, { "path": "api/migrations/0069_rename_to_in_trashcan.py", "content": "from django.db import migrations\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0068_remove_longrunningjob_result_and_more\"),\n ]\n\n operations = [\n migrations.RenameField(\n model_name=\"photo\",\n old_name=\"deleted\",\n new_name=\"in_trashcan\",\n ),\n ]\n" }, { "path": "api/migrations/0070_photo_removed.py", "content": "# Generated by Django 4.2.14 on 2024-08-21 19:17\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0069_rename_to_in_trashcan\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"photo\",\n name=\"removed\",\n field=models.BooleanField(db_index=True, default=False),\n ),\n ]\n" }, { "path": "api/migrations/0071_rename_person_label_probability_face_cluster_probability_and_more.py", "content": "# Generated by Django 4.2.16 on 2024-09-20 18:56\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0070_photo_removed\"),\n ]\n\n operations = [\n migrations.RenameField(\n model_name=\"face\",\n old_name=\"person_label_probability\",\n new_name=\"cluster_probability\",\n ),\n migrations.RemoveField(\n model_name=\"face\",\n name=\"person_label_is_inferred\",\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"classification_person\",\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"classification_faces\",\n to=\"api.person\",\n ),\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"classification_probability\",\n field=models.FloatField(db_index=True, default=0.0),\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"cluster_person\",\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"cluster_classification_faces\",\n to=\"api.person\",\n ),\n ),\n migrations.AddField(\n model_name=\"face\",\n name=\"deleted\",\n field=models.BooleanField(default=False),\n ),\n ]\n" }, { "path": "api/migrations/0072_alter_face_person.py", "content": "# Generated by Django 4.2.16 on 2024-09-20 19:07\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\n \"api\",\n \"0071_rename_person_label_probability_face_cluster_probability_and_more\",\n ),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"face\",\n name=\"person\",\n field=models.ForeignKey(\n null=True,\n on_delete=django.db.models.deletion.DO_NOTHING,\n related_name=\"faces\",\n to=\"api.person\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0073_remove_unknown_person.py", "content": "from django.db import migrations\n\n\ndef delete_unknown_person_and_update_faces(apps, schema_editor):\n # Get models\n Person = apps.get_model(\"api\", \"Person\")\n Face = apps.get_model(\"api\", \"Face\")\n\n # Define the name for unknown persons\n unknown_person_name = \"Unknown - Other\"\n\n # Find all persons with the name \"Unknown - Other\"\n unknown_persons = Person.objects.filter(name=unknown_person_name)\n\n # Iterate through each unknown person and set faces' person field to null\n for unknown_person in unknown_persons:\n # Set all faces' person field referencing the \"Unknown - Other\" person to null\n Face.objects.filter(person=unknown_person).update(person=None)\n\n # Delete the \"Unknown - Other\" person\n unknown_person.delete()\n\n\ndef recreate_unknown_person_and_restore_faces(apps, schema_editor):\n # Get models\n Person = apps.get_model(\"api\", \"Person\")\n Face = apps.get_model(\"api\", \"Face\")\n User = apps.get_model(\"api\", \"User\")\n\n # Define the name for unknown persons\n unknown_person_name = \"Unknown - Other\"\n\n # Retrieve all users to recreate their unknown persons\n users = User.objects.all()\n\n for user in users:\n # Recreate the \"Unknown - Other\" person for each user\n unknown_person = Person.objects.create(\n name=unknown_person_name, kind=Person.KIND_UNKNOWN, cluster_owner=user\n )\n\n # Restore faces for each recreated person based on user ownership\n Face.objects.filter(person=None, photo__owner=user).update(\n person=unknown_person\n )\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\n \"api\",\n \"0072_alter_face_person\",\n ),\n ]\n\n operations = [\n migrations.RunPython(\n delete_unknown_person_and_update_faces,\n recreate_unknown_person_and_restore_faces,\n ),\n ]\n" }, { "path": "api/migrations/0074_migrate_cluster_person.py", "content": "from django.db import migrations\n\n\ndef move_person_to_cluster_if_kind_cluster(apps, schema_editor):\n # Get the necessary models\n Face = apps.get_model(\"api\", \"Face\")\n\n # Define the constant for KIND_CLUSTER\n KIND_CLUSTER = \"CLUSTER\"\n\n # Fetch all Face instances where person is not null\n faces_to_update = Face.objects.filter(person__isnull=False)\n\n # Iterate over the faces and process each one\n for face in faces_to_update:\n # Check if the person is of type KIND_CLUSTER\n if face.person.kind == KIND_CLUSTER:\n # Move the person to the cluster field and set the person field to null\n face.cluster_person = face.person\n face.person = None\n face.save()\n\n\ndef restore_person_from_cluster(apps, schema_editor):\n # Get the necessary models\n Face = apps.get_model(\"api\", \"Face\")\n\n # Fetch all Face instances where original_person_id is not null (from forward migration)\n faces_to_restore = Face.objects.filter(cluster_person__isnull=False)\n\n # Iterate over the faces and restore the person reference from the original_person_id field\n for face in faces_to_restore:\n face.person = face.cluster_person\n face.cluster_person = None\n face.save()\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\n \"api\",\n \"0073_remove_unknown_person\",\n ),\n ]\n\n operations = [\n migrations.RunPython(\n move_person_to_cluster_if_kind_cluster,\n restore_person_from_cluster,\n ),\n ]\n" }, { "path": "api/migrations/0075_alter_face_cluster_person.py", "content": "# Generated by Django 4.2.16 on 2024-09-20 19:49\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0074_migrate_cluster_person\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"face\",\n name=\"cluster_person\",\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name=\"cluster_faces\",\n to=\"api.person\",\n ),\n ),\n ]\n" }, { "path": "api/migrations/0076_alter_file_path_alter_longrunningjob_job_type_and_more.py", "content": "# Generated by Django 4.2.18 on 2025-03-29 12:38\n\nfrom django.db import migrations, models\nimport django_cryptography.fields\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0075_alter_face_cluster_person\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"file\",\n name=\"path\",\n field=models.TextField(blank=True, default=\"\"),\n ),\n migrations.AlterField(\n model_name=\"longrunningjob\",\n name=\"job_type\",\n field=models.PositiveIntegerField(\n choices=[\n (1, \"Scan Photos\"),\n (2, \"Generate Event Albums\"),\n (3, \"Regenerate Event Titles\"),\n (4, \"Train Faces\"),\n (5, \"Delete Missing Photos\"),\n (7, \"Scan Faces\"),\n (6, \"Calculate Clip Embeddings\"),\n (8, \"Find Similar Faces\"),\n (9, \"Download Selected Photos\"),\n (10, \"Download Models\"),\n (11, \"Add Geolocation\"),\n (12, \"Generate Tags\"),\n (13, \"Generate Face Embeddings\"),\n (14, \"Scan Missing Photos\"),\n ]\n ),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_app_password\",\n field=django_cryptography.fields.encrypt(\n models.CharField(blank=True, default=\"\", max_length=64)\n ),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_scan_directory\",\n field=models.CharField(\n blank=True, db_index=True, default=\"\", max_length=512\n ),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_server_address\",\n field=models.CharField(blank=True, default=\"\", max_length=200),\n ),\n migrations.AlterField(\n model_name=\"user\",\n name=\"nextcloud_username\",\n field=models.CharField(blank=True, default=\"\", max_length=64),\n ),\n ]\n" }, { "path": "api/migrations/0077_alter_albumdate_title.py", "content": "# Generated by Django 4.2.18 on 2025-03-29 12:59\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0076_alter_file_path_alter_longrunningjob_job_type_and_more\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"albumdate\",\n name=\"title\",\n field=models.CharField(\n blank=True, db_index=True, default=\"\", max_length=512\n ),\n ),\n ]\n" }, { "path": "api/migrations/0078_create_photo_thumbnail.py", "content": "from django.db import migrations, models\nimport django.db.models.deletion\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n ('api', '0077_alter_albumdate_title'),\n ]\n\n operations = [\n migrations.CreateModel(\n name='Thumbnail',\n fields=[\n ('photo', models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, primary_key=True, related_name='thumbnail', serialize=False, to='api.photo')),\n ('thumbnail_big', models.ImageField(upload_to='thumbnails_big')),\n ('square_thumbnail', models.ImageField(upload_to='square_thumbnails')),\n ('square_thumbnail_small', models.ImageField(upload_to='square_thumbnails_small')),\n ('aspect_ratio', models.FloatField(blank=True, null=True)),\n ('dominant_color', models.TextField(blank=True, null=True)),\n ],\n ),\n migrations.RunSQL(\n sql=[\n \"\"\"\n INSERT INTO api_thumbnail (\n photo_id,\n thumbnail_big,\n square_thumbnail,\n square_thumbnail_small,\n aspect_ratio,\n dominant_color\n )\n SELECT \n image_hash,\n thumbnail_big,\n square_thumbnail,\n square_thumbnail_small,\n aspect_ratio,\n dominant_color\n FROM api_photo\n \"\"\"\n ],\n reverse_sql=[\n \"\"\"\n UPDATE api_photo p\n SET \n thumbnail_big = pt.thumbnail_big,\n square_thumbnail = pt.square_thumbnail,\n square_thumbnail_small = pt.square_thumbnail_small,\n aspect_ratio = pt.aspect_ratio,\n dominant_color = pt.dominant_color\n FROM api_thumbnail pt\n WHERE p.image_hash = pt.photo_id\n \"\"\"\n ]\n ),\n migrations.RemoveField(\n model_name='photo',\n name='aspect_ratio',\n ),\n migrations.RemoveField(\n model_name='photo',\n name='dominant_color',\n ),\n migrations.RemoveField(\n model_name='photo',\n name='square_thumbnail',\n ),\n migrations.RemoveField(\n model_name='photo',\n name='square_thumbnail_small',\n ),\n migrations.RemoveField(\n model_name='photo',\n name='thumbnail_big',\n ),\n ] " }, { "path": "api/migrations/0079_alter_albumauto_title.py", "content": "# Generated by Django 4.2.18 on 2025-05-04 14:28\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0078_create_photo_thumbnail\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"albumauto\",\n name=\"title\",\n field=models.CharField(default=\"Untitled Album\", max_length=512),\n ),\n ]\n" }, { "path": "api/migrations/0080_create_photo_caption.py", "content": "# Generated migration for PhotoCaption model\n\nfrom django.db import migrations, models\nimport django.db.models.deletion\n\n\ndef migrate_caption_data(apps, schema_editor):\n \"\"\"Migrate existing caption data from Photo to PhotoCaption\"\"\"\n Photo = apps.get_model('api', 'Photo')\n PhotoCaption = apps.get_model('api', 'PhotoCaption')\n \n db_alias = schema_editor.connection.alias\n \n # Create PhotoCaption instances for all photos that have caption data\n photos_with_captions = Photo.objects.using(db_alias).filter(\n models.Q(captions_json__isnull=False) | models.Q(search_captions__isnull=False)\n ).exclude(captions_json={})\n \n captions_to_create = []\n \n for photo in photos_with_captions.iterator():\n captions_to_create.append(\n PhotoCaption(\n photo_id=photo.image_hash,\n captions_json=photo.captions_json,\n search_captions=photo.search_captions\n )\n )\n \n # Process in batches to avoid memory issues\n if len(captions_to_create) >= 1000:\n PhotoCaption.objects.using(db_alias).bulk_create(captions_to_create, ignore_conflicts=True)\n captions_to_create = []\n \n # Create remaining captions\n if captions_to_create:\n PhotoCaption.objects.using(db_alias).bulk_create(captions_to_create, ignore_conflicts=True)\n\n\ndef reverse_migrate_caption_data(apps, schema_editor):\n \"\"\"Reverse migration - copy data back from PhotoCaption to Photo\"\"\"\n Photo = apps.get_model('api', 'Photo')\n PhotoCaption = apps.get_model('api', 'PhotoCaption')\n \n db_alias = schema_editor.connection.alias\n \n # Update photos with caption data from PhotoCaption instances\n for caption in PhotoCaption.objects.using(db_alias).all():\n try:\n photo = Photo.objects.using(db_alias).get(image_hash=caption.photo_id)\n photo.captions_json = caption.captions_json\n photo.search_captions = caption.search_captions\n photo.save(update_fields=['captions_json', 'search_captions'])\n except Photo.DoesNotExist:\n continue\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0079_alter_albumauto_title'),\n ]\n\n operations = [\n migrations.CreateModel(\n name='PhotoCaption',\n fields=[\n ('photo', models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, primary_key=True, related_name='caption_instance', serialize=False, to='api.photo')),\n ('captions_json', models.JSONField(blank=True, db_index=True, null=True)),\n ('search_captions', models.TextField(blank=True, db_index=True, null=True)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ],\n options={\n 'db_table': 'api_photo_caption',\n },\n ),\n migrations.RunPython(\n migrate_caption_data,\n reverse_migrate_caption_data,\n ),\n ] " }, { "path": "api/migrations/0081_remove_caption_fields_from_photo.py", "content": "# Generated migration to remove caption fields from Photo model\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0080_create_photo_caption'),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name='photo',\n name='captions_json',\n ),\n migrations.RemoveField(\n model_name='photo',\n name='search_captions',\n ),\n ] " }, { "path": "api/migrations/0082_create_photo_search.py", "content": "# Generated migration for PhotoSearch model\n\nfrom django.db import migrations, models\nimport django.db.models.deletion\n\n\ndef migrate_search_data(apps, schema_editor):\n \"\"\"Migrate existing search data from Photo and PhotoCaption to PhotoSearch\"\"\"\n Photo = apps.get_model('api', 'Photo')\n PhotoCaption = apps.get_model('api', 'PhotoCaption')\n PhotoSearch = apps.get_model('api', 'PhotoSearch')\n \n db_alias = schema_editor.connection.alias\n \n # Create PhotoSearch instances for all photos that have search data\n photos_with_search_data = Photo.objects.using(db_alias).filter(\n search_location__isnull=False\n ).exclude(search_location='')\n \n search_instances_to_create = []\n \n for photo in photos_with_search_data.iterator():\n search_instances_to_create.append(\n PhotoSearch(\n photo=photo,\n search_location=photo.search_location,\n search_captions='' # Will be populated later\n )\n )\n \n # Bulk create PhotoSearch instances\n PhotoSearch.objects.using(db_alias).bulk_create(search_instances_to_create, ignore_conflicts=True)\n \n # Now migrate search_captions from PhotoCaption to PhotoSearch\n for caption in PhotoCaption.objects.using(db_alias).filter(search_captions__isnull=False).exclude(search_captions=''):\n search_instance, created = PhotoSearch.objects.using(db_alias).get_or_create(\n photo=caption.photo,\n defaults={'search_captions': caption.search_captions, 'search_location': ''}\n )\n if not created:\n search_instance.search_captions = caption.search_captions\n search_instance.save()\n\n\ndef reverse_migrate_search_data(apps, schema_editor):\n \"\"\"Reverse migration - copy data back from PhotoSearch to Photo and PhotoCaption\"\"\"\n Photo = apps.get_model('api', 'Photo')\n PhotoCaption = apps.get_model('api', 'PhotoCaption')\n PhotoSearch = apps.get_model('api', 'PhotoSearch')\n \n db_alias = schema_editor.connection.alias\n \n # Update photos with search_location from PhotoSearch instances\n for search in PhotoSearch.objects.using(db_alias).all():\n try:\n photo = Photo.objects.using(db_alias).get(image_hash=search.photo_id)\n photo.search_location = search.search_location\n photo.save(update_fields=['search_location'])\n \n # Update PhotoCaption with search_captions\n caption, created = PhotoCaption.objects.using(db_alias).get_or_create(photo=photo)\n caption.search_captions = search.search_captions\n caption.save(update_fields=['search_captions'])\n except Photo.DoesNotExist:\n continue\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0081_remove_caption_fields_from_photo'),\n ]\n\n operations = [\n migrations.CreateModel(\n name='PhotoSearch',\n fields=[\n ('photo', models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, primary_key=True, related_name='search_instance', serialize=False, to='api.photo')),\n ('search_captions', models.TextField(blank=True, db_index=True, null=True)),\n ('search_location', models.TextField(blank=True, db_index=True, null=True)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ],\n options={\n 'db_table': 'api_photo_search',\n },\n ),\n migrations.RunPython(\n migrate_search_data,\n reverse_migrate_search_data,\n ),\n ] " }, { "path": "api/migrations/0083_remove_search_fields.py", "content": "# Generated migration to remove search fields from Photo and PhotoCaption models\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0082_create_photo_search'),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name='photo',\n name='search_location',\n ),\n migrations.RemoveField(\n model_name='photocaption',\n name='search_captions',\n ),\n ] " }, { "path": "api/migrations/0084_convert_arrayfield_to_json.py", "content": "# Migration to safely convert ArrayField to JSONField for SQLite compatibility\nfrom django.db import migrations, models\n\n\ndef copy_arrayfield_to_json(apps, schema_editor):\n \"\"\"\n Copy data from ArrayField to JSONField format.\n This handles the conversion for both PostgreSQL and SQLite.\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n \n for photo in Photo.objects.all():\n if photo.clip_embeddings is not None:\n # ArrayField data is already in list format, just copy it\n photo.clip_embeddings_json = photo.clip_embeddings\n photo.save(update_fields=['clip_embeddings_json'])\n\n\ndef copy_json_to_arrayfield(apps, schema_editor):\n \"\"\"\n Reverse migration: copy JSONField data back to ArrayField format.\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n \n for photo in Photo.objects.all():\n if photo.clip_embeddings_json is not None:\n photo.clip_embeddings = photo.clip_embeddings_json\n photo.save(update_fields=['clip_embeddings'])\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n ('api', '0083_remove_search_fields'),\n ]\n\n operations = [\n # Step 1: Add new JSONField\n migrations.AddField(\n model_name='Photo',\n name='clip_embeddings_json',\n field=models.JSONField(blank=True, null=True),\n ),\n \n # Step 2: Copy data from ArrayField to JSONField\n migrations.RunPython(\n copy_arrayfield_to_json,\n copy_json_to_arrayfield,\n ),\n \n # Step 3: Remove old ArrayField\n migrations.RemoveField(\n model_name='Photo',\n name='clip_embeddings',\n ),\n \n # Step 4: Rename JSONField to original name\n migrations.RenameField(\n model_name='Photo',\n old_name='clip_embeddings_json',\n new_name='clip_embeddings',\n ),\n ] " }, { "path": "api/migrations/0085_albumuser_public_expires_at_albumuser_public_slug.py", "content": "# Generated by Django 5.2.4 on 2025-08-16 08:40\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0084_convert_arrayfield_to_json'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='albumuser',\n name='public_expires_at',\n field=models.DateTimeField(blank=True, db_index=True, null=True),\n ),\n migrations.AddField(\n model_name='albumuser',\n name='public_slug',\n field=models.SlugField(blank=True, max_length=64, null=True, unique=True),\n ),\n ]\n" }, { "path": "api/migrations/0086_remove_albumuser_public_and_more.py", "content": "# Generated by Django 5.2.4 on 2025-08-17 17:23\n\nimport django.db.models.deletion\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0085_albumuser_public_expires_at_albumuser_public_slug'),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name='albumuser',\n name='public',\n ),\n migrations.RemoveField(\n model_name='albumuser',\n name='public_expires_at',\n ),\n migrations.RemoveField(\n model_name='albumuser',\n name='public_slug',\n ),\n migrations.CreateModel(\n name='AlbumUserShare',\n fields=[\n ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),\n ('enabled', models.BooleanField(db_index=True, default=False)),\n ('slug', models.SlugField(blank=True, max_length=64, null=True, unique=True)),\n ('expires_at', models.DateTimeField(blank=True, db_index=True, null=True)),\n ('album', models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, related_name='share', to='api.albumuser')),\n ],\n ),\n ]\n" }, { "path": "api/migrations/0087_add_folder_album.py", "content": "# Generated by Django 5.2.4 on 2025-08-22 14:48\n\nimport django.db.models.deletion\nfrom django.conf import settings\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0086_remove_albumuser_public_and_more'),\n ]\n\n operations = [\n migrations.CreateModel(\n name='FolderAlbum',\n fields=[\n ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),\n ('title', models.CharField(max_length=512)),\n ('created_on', models.DateTimeField(auto_now_add=True, db_index=True)),\n ('updated_on', models.DateTimeField(auto_now=True)),\n ('folder_path', models.TextField()),\n ('include_subdirectories', models.BooleanField(default=True)),\n ('public', models.BooleanField(db_index=True, default=False)),\n ('favorited', models.BooleanField(db_index=True, default=False)),\n ('cover_photo', models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='folder_album_cover', to='api.photo')),\n ('owner', models.ForeignKey(default=None, on_delete=django.db.models.deletion.CASCADE, to=settings.AUTH_USER_MODEL)),\n ],\n options={\n 'ordering': ['-created_on'],\n 'unique_together': {('folder_path', 'owner')},\n },\n ),\n ]\n" }, { "path": "api/migrations/0088_remove_folder_album.py", "content": "# Generated by Django 5.2.4 on 2025-08-22 15:40\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0087_add_folder_album'),\n ]\n\n operations = [\n migrations.DeleteModel(\n name='FolderAlbum',\n ),\n ]\n" }, { "path": "api/migrations/0089_add_text_alignment.py", "content": "# Generated by Django 5.2.4 on 2025-08-22 16:00\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0088_remove_folder_album'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='user',\n name='text_alignment',\n field=models.TextField(choices=[('left', 'Left'), ('right', 'Right')], default='right'),\n ),\n ]\n" }, { "path": "api/migrations/0090_add_header_size.py", "content": "# Generated by Django 5.2.4 on 2025-08-22 16:27\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0089_add_text_alignment'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='user',\n name='header_size',\n field=models.TextField(choices=[('large', 'Large'), ('normal', 'Normal'), ('small', 'Small')], default='large'),\n ),\n ]\n" }, { "path": "api/migrations/0091_alter_user_scan_directory.py", "content": "# Generated by Django 5.2.4 on 2025-08-30 12:34\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0090_add_header_size'),\n ]\n\n operations = [\n migrations.AlterField(\n model_name='user',\n name='scan_directory',\n field=models.CharField(blank=True, db_index=True, default='', max_length=512),\n ),\n ]\n" }, { "path": "api/migrations/0092_add_skip_raw_files_field.py", "content": "# Generated by Django 5.2.7 on 2025-10-26 21:54\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0091_alter_user_scan_directory'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='user',\n name='skip_raw_files',\n field=models.BooleanField(default=False),\n ),\n ]\n" }, { "path": "api/migrations/0093_migrate_photon_to_nominatim.py", "content": "\"\"\"\nMigration to change MAP_API_PROVIDER from 'photon' to 'nominatim'.\n\nPhoton's public API at photon.komoot.io has become unreliable (502 errors),\nso we're switching the default to Nominatim which is more stable.\n\"\"\"\n\nfrom django.db import migrations\n\n\ndef migrate_photon_to_nominatim(apps, schema_editor):\n \"\"\"Update constance config from photon to nominatim.\"\"\"\n try:\n Constance = apps.get_model(\"constance\", \"Constance\")\n config = Constance.objects.filter(key=\"MAP_API_PROVIDER\").first()\n if config and config.value == '\"photon\"':\n config.value = '\"nominatim\"'\n config.save()\n except LookupError:\n # constance model not available, skip\n pass\n\n\ndef reverse_migration(apps, schema_editor):\n \"\"\"Reverse: change nominatim back to photon (not recommended).\"\"\"\n try:\n Constance = apps.get_model(\"constance\", \"Constance\")\n config = Constance.objects.filter(key=\"MAP_API_PROVIDER\").first()\n if config and config.value == '\"nominatim\"':\n config.value = '\"photon\"'\n config.save()\n except LookupError:\n pass\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0092_add_skip_raw_files_field\"),\n ]\n\n operations = [\n migrations.RunPython(migrate_photon_to_nominatim, reverse_migration),\n ]\n\n" }, { "path": "api/migrations/0094_add_slideshow_interval.py", "content": "# Generated manually for slideshow interval feature\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0093_migrate_photon_to_nominatim'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='user',\n name='slideshow_interval',\n field=models.IntegerField(default=5),\n ),\n ]\n\n" }, { "path": "api/migrations/0095_photo_perceptual_hash_alter_longrunningjob_job_type_and_more.py", "content": "# Generated by Django 5.2.9 on 2025-12-23 09:47\n\nimport api.models.user\nimport django.db.models.deletion\nfrom django.conf import settings\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0094_add_slideshow_interval'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='photo',\n name='perceptual_hash',\n field=models.CharField(blank=True, db_index=True, max_length=64, null=True),\n ),\n migrations.AlterField(\n model_name='longrunningjob',\n name='job_type',\n field=models.PositiveIntegerField(choices=[(1, 'Scan Photos'), (2, 'Generate Event Albums'), (3, 'Regenerate Event Titles'), (4, 'Train Faces'), (5, 'Delete Missing Photos'), (7, 'Scan Faces'), (6, 'Calculate Clip Embeddings'), (8, 'Find Similar Faces'), (9, 'Download Selected Photos'), (10, 'Download Models'), (11, 'Add Geolocation'), (12, 'Generate Tags'), (13, 'Generate Face Embeddings'), (14, 'Scan Missing Photos'), (15, 'Detect Duplicate Photos')]),\n ),\n migrations.CreateModel(\n name='DuplicateGroup',\n fields=[\n ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ('status', models.CharField(choices=[('pending', 'Pending Review'), ('reviewed', 'Reviewed'), ('dismissed', 'Dismissed (Not Duplicates)')], db_index=True, default='pending', max_length=20)),\n ('owner', models.ForeignKey(on_delete=models.SET(api.models.user.get_deleted_user), related_name='duplicate_groups', to=settings.AUTH_USER_MODEL)),\n ('preferred_photo', models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='preferred_in_group', to='api.photo')),\n ],\n options={\n 'ordering': ['-created_at'],\n },\n ),\n migrations.AddField(\n model_name='photo',\n name='duplicate_group',\n field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='photos', to='api.duplicategroup'),\n ),\n ]\n" }, { "path": "api/migrations/0096_add_progress_step_and_result_to_longrunningjob.py", "content": "# Generated by Django 5.2.9 on 2025-12-23 12:13\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0095_photo_perceptual_hash_alter_longrunningjob_job_type_and_more'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='longrunningjob',\n name='progress_step',\n field=models.CharField(blank=True, max_length=100, null=True),\n ),\n migrations.AddField(\n model_name='longrunningjob',\n name='result',\n field=models.JSONField(blank=True, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0097_add_duplicate_detection_settings_to_user.py", "content": "# Generated by Django 5.2.9 on 2025-12-23 13:00\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0096_add_progress_step_and_result_to_longrunningjob'),\n ]\n\n operations = [\n migrations.AddField(\n model_name='user',\n name='duplicate_clear_existing',\n field=models.BooleanField(default=False),\n ),\n migrations.AddField(\n model_name='user',\n name='duplicate_sensitivity',\n field=models.TextField(choices=[('strict', 'Strict'), ('normal', 'Normal'), ('loose', 'Loose')], default='normal'),\n ),\n ]\n" }, { "path": "api/migrations/0098_add_photo_stack.py", "content": "# Generated migration for PhotoStack unified grouping system\n\nimport uuid\nfrom django.db import migrations, models\nimport django.db.models.deletion\n\n\ndef get_deleted_user():\n \"\"\"Reference to the function that returns the deleted user placeholder.\"\"\"\n from api.models.user import get_deleted_user as _get_deleted_user\n return _get_deleted_user\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0097_add_duplicate_detection_settings_to_user'),\n ]\n\n operations = [\n # Create the PhotoStack table\n migrations.CreateModel(\n name='PhotoStack',\n fields=[\n ('id', models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False)),\n ('stack_type', models.CharField(\n choices=[\n ('exact_copy', 'Exact Copies'),\n ('visual_duplicate', 'Visual Duplicates'),\n ('raw_jpeg', 'RAW + JPEG Pair'),\n ('burst', 'Burst Sequence'),\n ('bracket', 'Exposure Bracket'),\n ('live_photo', 'Live Photo'),\n ('manual', 'Manual Stack'),\n ],\n db_index=True,\n default='visual_duplicate',\n max_length=20,\n )),\n ('status', models.CharField(\n choices=[\n ('pending', 'Pending Review'),\n ('reviewed', 'Reviewed'),\n ('dismissed', 'Dismissed'),\n ],\n db_index=True,\n default='pending',\n max_length=20,\n )),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ('similarity_score', models.FloatField(blank=True, null=True)),\n ('sequence_start', models.DateTimeField(blank=True, null=True)),\n ('sequence_end', models.DateTimeField(blank=True, null=True)),\n ('potential_savings', models.BigIntegerField(default=0)),\n ('owner', models.ForeignKey(\n on_delete=django.db.models.deletion.SET(get_deleted_user),\n related_name='photo_stacks',\n to='api.user',\n )),\n ('primary_photo', models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name='primary_in_stack',\n to='api.photo',\n )),\n ],\n options={\n 'verbose_name': 'Photo Stack',\n 'verbose_name_plural': 'Photo Stacks',\n 'ordering': ['-created_at'],\n },\n ),\n # Add indexes for PhotoStack\n migrations.AddIndex(\n model_name='photostack',\n index=models.Index(fields=['owner', 'stack_type', 'status'], name='api_photost_owner_i_abc123_idx'),\n ),\n migrations.AddIndex(\n model_name='photostack',\n index=models.Index(fields=['owner', 'status'], name='api_photost_owner_i_def456_idx'),\n ),\n # Add stack field to Photo\n migrations.AddField(\n model_name='photo',\n name='stack',\n field=models.ForeignKey(\n blank=True,\n null=True,\n on_delete=django.db.models.deletion.SET_NULL,\n related_name='photos',\n to='api.photostack',\n ),\n ),\n # Add sub-second timestamp for burst detection\n migrations.AddField(\n model_name='photo',\n name='exif_timestamp_subsec',\n field=models.CharField(blank=True, max_length=10, null=True),\n ),\n # Add image sequence number for burst detection\n migrations.AddField(\n model_name='photo',\n name='image_sequence_number',\n field=models.IntegerField(blank=True, null=True),\n ),\n # Remove duplicate_group FK from Photo (replaced by stack)\n migrations.RemoveField(\n model_name='photo',\n name='duplicate_group',\n ),\n # Delete the old DuplicateGroup model\n migrations.DeleteModel(\n name='DuplicateGroup',\n ),\n ]\n" }, { "path": "api/migrations/0099_photo_uuid_primary_key.py", "content": "# Generated migration for Photo UUID primary key\n# This migration changes Photo from using image_hash as PK to using UUID as PK\n#\n# Supports both PostgreSQL (raw SQL) and SQLite (table recreation pattern).\n#\n# ============================================================================\n# CRITICAL WARNING: THIS MIGRATION IS NOT REVERSIBLE\n# ============================================================================\n# This migration fundamentally changes the Photo primary key from image_hash\n# (a content-based hash) to UUID (a random identifier). Reversing this would\n# require regenerating the original image_hash values from file content, which\n# is not possible without access to the original photo files and significant\n# processing time.\n#\n# BEFORE RUNNING THIS MIGRATION:\n# 1. Create a FULL DATABASE BACKUP: pg_dump -U your_user your_db > backup.sql\n# 2. Test the migration on a copy of your production database first\n# 3. Plan for downtime - this migration may take significant time on large DBs\n# 4. Ensure you have enough disk space for the migration operations\n#\n# TO ROLLBACK (if needed):\n# 1. Stop the application\n# 2. Restore from your pre-migration database backup\n# 3. Fake-migrate back: python manage.py migrate api 0098 --fake\n# ============================================================================\n\nimport uuid\nfrom django.db import migrations, models\n\n\n# ============================================================================\n# PostgreSQL raw SQL\n# ============================================================================\n\nPOSTGRESQL_FORWARD_SQL = \"\"\"\n-- Step 1: Add UUID column to api_photo\nALTER TABLE api_photo ADD COLUMN id UUID DEFAULT gen_random_uuid();\nUPDATE api_photo SET id = gen_random_uuid() WHERE id IS NULL;\nALTER TABLE api_photo ALTER COLUMN id SET NOT NULL;\n\n-- Step 2: Create a mapping table for old hash -> new UUID\nCREATE TEMP TABLE photo_id_mapping AS\nSELECT image_hash, id FROM api_photo;\nCREATE INDEX ON photo_id_mapping(image_hash);\n\n-- Step 3: Add new UUID columns to all related tables\n\n-- api_face\nALTER TABLE api_face ADD COLUMN photo_id_new UUID;\nUPDATE api_face f SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE f.photo_id = m.image_hash;\n\n-- api_photo_shared_to (M2M through table)\nALTER TABLE api_photo_shared_to ADD COLUMN photo_id_new UUID;\nUPDATE api_photo_shared_to t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_photo_files (M2M through table)\nALTER TABLE api_photo_files ADD COLUMN photo_id_new UUID;\nUPDATE api_photo_files t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_albumuser_photos (M2M through table)\nALTER TABLE api_albumuser_photos ADD COLUMN photo_id_new UUID;\nUPDATE api_albumuser_photos t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_albumthing_photos (M2M through table)\nALTER TABLE api_albumthing_photos ADD COLUMN photo_id_new UUID;\nUPDATE api_albumthing_photos t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_albumplace_photos (M2M through table)\nALTER TABLE api_albumplace_photos ADD COLUMN photo_id_new UUID;\nUPDATE api_albumplace_photos t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_albumdate_photos (M2M through table)\nALTER TABLE api_albumdate_photos ADD COLUMN photo_id_new UUID;\nUPDATE api_albumdate_photos t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_albumauto_photos (M2M through table)\nALTER TABLE api_albumauto_photos ADD COLUMN photo_id_new UUID;\nUPDATE api_albumauto_photos t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_person (cover_photo_id)\nALTER TABLE api_person ADD COLUMN cover_photo_id_new UUID;\nUPDATE api_person p SET cover_photo_id_new = m.id\nFROM photo_id_mapping m WHERE p.cover_photo_id = m.image_hash;\n\n-- api_albumuser (cover_photo_id)\nALTER TABLE api_albumuser ADD COLUMN cover_photo_id_new UUID;\nUPDATE api_albumuser a SET cover_photo_id_new = m.id\nFROM photo_id_mapping m WHERE a.cover_photo_id = m.image_hash;\n\n-- api_albumthing_cover_photos (M2M through table for cover photos)\nALTER TABLE api_albumthing_cover_photos ADD COLUMN photo_id_new UUID;\nUPDATE api_albumthing_cover_photos t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_thumbnail (OneToOne with PK)\nALTER TABLE api_thumbnail ADD COLUMN photo_id_new UUID;\nUPDATE api_thumbnail t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_photo_caption (OneToOne with PK)\nALTER TABLE api_photo_caption ADD COLUMN photo_id_new UUID;\nUPDATE api_photo_caption t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_photo_search (OneToOne with PK)\nALTER TABLE api_photo_search ADD COLUMN photo_id_new UUID;\nUPDATE api_photo_search t SET photo_id_new = m.id\nFROM photo_id_mapping m WHERE t.photo_id = m.image_hash;\n\n-- api_photostack (primary_photo_id)\nALTER TABLE api_photostack ADD COLUMN primary_photo_id_new UUID;\nUPDATE api_photostack s SET primary_photo_id_new = m.id\nFROM photo_id_mapping m WHERE s.primary_photo_id = m.image_hash;\n\n-- Step 4: Drop all FK constraints\nALTER TABLE api_face DROP CONSTRAINT IF EXISTS api_face_photo_id_6f997226_fk_api_photo_image_hash;\nALTER TABLE api_photo_shared_to DROP CONSTRAINT IF EXISTS api_photo_shared_to_photo_id_852923c7_fk_api_photo_image_hash;\nALTER TABLE api_photo_files DROP CONSTRAINT IF EXISTS api_photo_files_photo_id_f4365127_fk_api_photo_image_hash;\nALTER TABLE api_albumuser_photos DROP CONSTRAINT IF EXISTS api_albumuser_photos_photo_id_b9df1b14_fk_api_photo_image_hash;\nALTER TABLE api_albumthing_photos DROP CONSTRAINT IF EXISTS api_albumthing_photos_photo_id_d0832fc3_fk_api_photo_image_hash;\nALTER TABLE api_albumplace_photos DROP CONSTRAINT IF EXISTS api_albumplace_photos_photo_id_8fd88190_fk_api_photo_image_hash;\nALTER TABLE api_albumdate_photos DROP CONSTRAINT IF EXISTS api_albumdate_photos_photo_id_26095959_fk_api_photo_image_hash;\nALTER TABLE api_albumauto_photos DROP CONSTRAINT IF EXISTS api_albumauto_photos_photo_id_3320c2f0_fk_api_photo_image_hash;\nALTER TABLE api_person DROP CONSTRAINT IF EXISTS api_person_cover_photo_id_e0d8a6ab_fk_api_photo_image_hash;\nALTER TABLE api_albumuser DROP CONSTRAINT IF EXISTS api_albumuser_cover_photo_id_69b304ac_fk_api_photo_image_hash;\nALTER TABLE api_albumthing_cover_photos DROP CONSTRAINT IF EXISTS api_albumthing_cover_photo_id_ae113997_fk_api_photo;\nALTER TABLE api_thumbnail DROP CONSTRAINT IF EXISTS api_thumbnail_photo_id_484afcd0_fk_api_photo_image_hash;\nALTER TABLE api_photo_caption DROP CONSTRAINT IF EXISTS api_photo_caption_photo_id_363f8856_fk_api_photo_image_hash;\nALTER TABLE api_photo_search DROP CONSTRAINT IF EXISTS api_photo_search_photo_id_b4055a77_fk_api_photo_image_hash;\nALTER TABLE api_photostack DROP CONSTRAINT IF EXISTS api_photostack_primary_photo_id_a2e9fc96_fk_api_photo;\n\n-- Step 5: Drop PKs on related tables that use photo as PK\nALTER TABLE api_thumbnail DROP CONSTRAINT IF EXISTS api_thumbnail_pkey;\nALTER TABLE api_photo_caption DROP CONSTRAINT IF EXISTS api_photo_caption_pkey;\nALTER TABLE api_photo_search DROP CONSTRAINT IF EXISTS api_photo_search_pkey;\n\n-- Step 6: Drop old PK on api_photo, add new one\nALTER TABLE api_photo DROP CONSTRAINT api_photo_pkey;\nALTER TABLE api_photo ADD PRIMARY KEY (id);\n\n-- Step 7: Add unique constraint on image_hash (for deduplication)\nCREATE UNIQUE INDEX api_photo_image_hash_unique ON api_photo(image_hash);\n\n-- Step 8: Drop old FK columns, rename new ones\n\nALTER TABLE api_face DROP COLUMN photo_id;\nALTER TABLE api_face RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_photo_shared_to DROP COLUMN photo_id;\nALTER TABLE api_photo_shared_to RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_photo_files DROP COLUMN photo_id;\nALTER TABLE api_photo_files RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_albumuser_photos DROP COLUMN photo_id;\nALTER TABLE api_albumuser_photos RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_albumthing_photos DROP COLUMN photo_id;\nALTER TABLE api_albumthing_photos RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_albumplace_photos DROP COLUMN photo_id;\nALTER TABLE api_albumplace_photos RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_albumdate_photos DROP COLUMN photo_id;\nALTER TABLE api_albumdate_photos RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_albumauto_photos DROP COLUMN photo_id;\nALTER TABLE api_albumauto_photos RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_person DROP COLUMN cover_photo_id;\nALTER TABLE api_person RENAME COLUMN cover_photo_id_new TO cover_photo_id;\n\nALTER TABLE api_albumuser DROP COLUMN cover_photo_id;\nALTER TABLE api_albumuser RENAME COLUMN cover_photo_id_new TO cover_photo_id;\n\nALTER TABLE api_albumthing_cover_photos DROP COLUMN photo_id;\nALTER TABLE api_albumthing_cover_photos RENAME COLUMN photo_id_new TO photo_id;\n\nALTER TABLE api_thumbnail DROP COLUMN photo_id;\nALTER TABLE api_thumbnail RENAME COLUMN photo_id_new TO photo_id;\nALTER TABLE api_thumbnail ALTER COLUMN photo_id SET NOT NULL;\nALTER TABLE api_thumbnail ADD PRIMARY KEY (photo_id);\n\nALTER TABLE api_photo_caption DROP COLUMN photo_id;\nALTER TABLE api_photo_caption RENAME COLUMN photo_id_new TO photo_id;\nALTER TABLE api_photo_caption ALTER COLUMN photo_id SET NOT NULL;\nALTER TABLE api_photo_caption ADD PRIMARY KEY (photo_id);\n\nALTER TABLE api_photo_search DROP COLUMN photo_id;\nALTER TABLE api_photo_search RENAME COLUMN photo_id_new TO photo_id;\nALTER TABLE api_photo_search ALTER COLUMN photo_id SET NOT NULL;\nALTER TABLE api_photo_search ADD PRIMARY KEY (photo_id);\n\nALTER TABLE api_photostack DROP COLUMN primary_photo_id;\nALTER TABLE api_photostack RENAME COLUMN primary_photo_id_new TO primary_photo_id;\n\n-- Step 9: Recreate all FK constraints with new UUID type\nALTER TABLE api_face ADD CONSTRAINT api_face_photo_id_fk_api_photo\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_photo_shared_to ADD CONSTRAINT api_photo_shared_to_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_photo_files ADD CONSTRAINT api_photo_files_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_albumuser_photos ADD CONSTRAINT api_albumuser_photos_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_albumthing_photos ADD CONSTRAINT api_albumthing_photos_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_albumplace_photos ADD CONSTRAINT api_albumplace_photos_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_albumdate_photos ADD CONSTRAINT api_albumdate_photos_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_albumauto_photos ADD CONSTRAINT api_albumauto_photos_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_person ADD CONSTRAINT api_person_cover_photo_id_fk\n FOREIGN KEY (cover_photo_id) REFERENCES api_photo(id) ON DELETE SET NULL;\n\nALTER TABLE api_albumuser ADD CONSTRAINT api_albumuser_cover_photo_id_fk\n FOREIGN KEY (cover_photo_id) REFERENCES api_photo(id) ON DELETE SET NULL;\n\nALTER TABLE api_albumthing_cover_photos ADD CONSTRAINT api_albumthing_cover_photos_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_thumbnail ADD CONSTRAINT api_thumbnail_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_photo_caption ADD CONSTRAINT api_photo_caption_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_photo_search ADD CONSTRAINT api_photo_search_photo_id_fk\n FOREIGN KEY (photo_id) REFERENCES api_photo(id) ON DELETE CASCADE;\n\nALTER TABLE api_photostack ADD CONSTRAINT api_photostack_primary_photo_id_fk\n FOREIGN KEY (primary_photo_id) REFERENCES api_photo(id) ON DELETE SET NULL;\n\n-- Step 10: Create indexes for performance\nCREATE INDEX api_face_photo_id_idx ON api_face(photo_id);\nCREATE INDEX api_photo_shared_to_photo_id_idx ON api_photo_shared_to(photo_id);\nCREATE INDEX api_photo_files_photo_id_idx ON api_photo_files(photo_id);\nCREATE INDEX api_person_cover_photo_id_idx ON api_person(cover_photo_id);\nCREATE INDEX api_albumuser_cover_photo_id_idx ON api_albumuser(cover_photo_id);\nCREATE INDEX api_photostack_primary_photo_id_idx ON api_photostack(primary_photo_id);\n\n-- Clean up temp table\nDROP TABLE photo_id_mapping;\n\"\"\"\n\n\n# ============================================================================\n# Forward / reverse Python entry points\n# ============================================================================\n\ndef migrate_forward(apps, schema_editor):\n \"\"\"Forward migration - dispatches to PostgreSQL or SQLite implementation.\"\"\"\n vendor = schema_editor.connection.vendor\n if vendor == \"postgresql\":\n _migrate_postgresql(schema_editor)\n elif vendor == \"sqlite\":\n _migrate_sqlite(schema_editor)\n else:\n raise ValueError(\n f\"Unsupported database backend: {vendor}. \"\n f\"This migration supports PostgreSQL and SQLite only.\"\n )\n\n\ndef migrate_reverse(apps, schema_editor):\n \"\"\"Reverse migration - not supported for any backend.\"\"\"\n raise RuntimeError(\n \"Migration 0099_photo_uuid_primary_key cannot be reversed automatically. \"\n \"Please restore from your pre-migration database backup and run: \"\n \"python manage.py migrate api 0098 --fake\"\n )\n\n\n# ============================================================================\n# PostgreSQL implementation\n# ============================================================================\n\ndef _migrate_postgresql(schema_editor):\n \"\"\"Execute the PostgreSQL-specific migration using raw SQL.\"\"\"\n statements = schema_editor.connection.ops.prepare_sql_script(\n POSTGRESQL_FORWARD_SQL\n )\n for statement in statements:\n schema_editor.execute(statement)\n\n\n# ============================================================================\n# SQLite implementation (table-recreation pattern)\n# ============================================================================\n#\n# SQLite does not support most ALTER TABLE operations required by the\n# PostgreSQL path (DROP/ADD CONSTRAINT, ADD PRIMARY KEY, ALTER COLUMN,\n# UPDATE … FROM, gen_random_uuid(), etc.).\n#\n# Instead we use the standard SQLite table-recreation pattern:\n# 1. CREATE TABLE …__new (with the desired schema)\n# 2. INSERT INTO …__new SELECT … FROM … (copy data, translating FKs)\n# 3. DROP TABLE …\n# 4. ALTER TABLE …__new RENAME TO …\n# 5. Re-create indexes\n# ============================================================================\n\ndef _migrate_sqlite(schema_editor):\n \"\"\"Execute the SQLite-compatible migration via table recreation.\"\"\"\n cursor = schema_editor.connection.cursor()\n\n # Disable FK enforcement while we recreate tables\n cursor.execute(\"PRAGMA foreign_keys = OFF\")\n\n try:\n # -- Step 1: Build image_hash → UUID mapping --------------------------\n cursor.execute('SELECT \"image_hash\" FROM \"api_photo\"')\n mapping = {row[0]: str(uuid.uuid4()) for row in cursor.fetchall()}\n\n # -- Step 2: Add id column to api_photo and populate UUIDs ------------\n cursor.execute('ALTER TABLE \"api_photo\" ADD COLUMN \"id\" TEXT')\n for image_hash, new_id in mapping.items():\n cursor.execute(\n 'UPDATE \"api_photo\" SET \"id\" = ? WHERE \"image_hash\" = ?',\n [new_id, image_hash],\n )\n\n # -- Step 3: Recreate api_photo with id as PK ------------------------\n _sqlite_recreate_table(\n cursor,\n \"api_photo\",\n pk_column=\"id\",\n column_overrides={\n \"id\": '\"id\" TEXT NOT NULL',\n \"image_hash\": '\"image_hash\" varchar(64) NOT NULL UNIQUE',\n },\n )\n\n # -- Step 4: Update FK references in every related table --------------\n _FK_TABLES = [\n (\"api_face\", \"photo_id\"),\n (\"api_photo_shared_to\", \"photo_id\"),\n (\"api_photo_files\", \"photo_id\"),\n (\"api_albumuser_photos\", \"photo_id\"),\n (\"api_albumthing_photos\", \"photo_id\"),\n (\"api_albumplace_photos\", \"photo_id\"),\n (\"api_albumdate_photos\", \"photo_id\"),\n (\"api_albumauto_photos\", \"photo_id\"),\n (\"api_albumthing_cover_photos\", \"photo_id\"),\n (\"api_person\", \"cover_photo_id\"),\n (\"api_albumuser\", \"cover_photo_id\"),\n (\"api_photostack\", \"primary_photo_id\"),\n (\"api_thumbnail\", \"photo_id\"),\n (\"api_photo_caption\", \"photo_id\"),\n (\"api_photo_search\", \"photo_id\"),\n ]\n for table_name, fk_column in _FK_TABLES:\n _sqlite_update_fk_table(cursor, table_name, fk_column, mapping)\n\n # -- Step 5: Create performance indexes -------------------------------\n _sqlite_create_indexes(cursor)\n\n finally:\n cursor.execute(\"PRAGMA foreign_keys = ON\")\n\n\n# -- SQLite helpers -----------------------------------------------------------\n\ndef _sqlite_recreate_table(cursor, table_name, pk_column, column_overrides):\n \"\"\"\n Recreate *api_photo* with a new primary-key column.\n\n * ``pk_column`` – name of the column that becomes PRIMARY KEY\n * ``column_overrides`` – dict column_name → SQL column-definition fragment\n (without PRIMARY KEY – that is appended automatically\n for *pk_column*)\n \"\"\"\n columns = _sqlite_column_info(cursor, table_name)\n indexes = _sqlite_index_info(cursor, table_name)\n\n col_defs = []\n col_names = []\n for _cid, name, type_, notnull, dflt_value, _pk in columns:\n col_names.append(name)\n if name in column_overrides:\n defn = column_overrides[name]\n if name == pk_column:\n defn += \" PRIMARY KEY\"\n col_defs.append(defn)\n else:\n parts = [f'\"{name}\"', type_ or \"TEXT\"]\n if name == pk_column:\n parts.append(\"NOT NULL PRIMARY KEY\")\n elif notnull:\n parts.append(\"NOT NULL\")\n if dflt_value is not None and name != pk_column:\n parts.append(f\"DEFAULT {dflt_value}\")\n col_defs.append(\" \".join(parts))\n\n cols_quoted = \", \".join(f'\"{c}\"' for c in col_names)\n new_table = f\"{table_name}__new\"\n\n cursor.execute(f'CREATE TABLE \"{new_table}\" ({\", \".join(col_defs)})')\n cursor.execute(\n f'INSERT INTO \"{new_table}\" ({cols_quoted}) '\n f'SELECT {cols_quoted} FROM \"{table_name}\"'\n )\n cursor.execute(f'DROP TABLE \"{table_name}\"')\n cursor.execute(f'ALTER TABLE \"{new_table}\" RENAME TO \"{table_name}\"')\n\n # Re-create any existing indexes\n for _idx_name, idx_sql in indexes:\n try:\n cursor.execute(idx_sql)\n except Exception:\n pass # index may conflict with new UNIQUE constraint\n\n\ndef _sqlite_update_fk_table(cursor, table_name, fk_column, mapping):\n \"\"\"\n Recreate *table_name* so that every value in *fk_column* is translated\n from the old image_hash to the new UUID via *mapping*.\n \"\"\"\n # Guard: skip if the table or column doesn't exist\n cursor.execute(\n \"SELECT COUNT(*) FROM sqlite_master WHERE type='table' AND name=?\",\n [table_name],\n )\n if cursor.fetchone()[0] == 0:\n return\n\n columns = _sqlite_column_info(cursor, table_name)\n col_names_list = [col[1] for col in columns]\n if fk_column not in col_names_list:\n return\n\n indexes = _sqlite_index_info(cursor, table_name)\n\n # Build column definitions – change the FK column type to TEXT\n col_defs = []\n col_names = []\n fk_col_idx = None\n\n for _cid, name, type_, notnull, dflt_value, pk in columns:\n col_names.append(name)\n if name == fk_column:\n fk_col_idx = len(col_names) - 1\n parts = [f'\"{name}\" TEXT']\n if pk:\n parts.append(\"NOT NULL PRIMARY KEY\")\n col_defs.append(\" \".join(parts))\n else:\n parts = [f'\"{name}\"', type_ or \"TEXT\"]\n if pk:\n parts.append(\"PRIMARY KEY\")\n if notnull and not pk:\n parts.append(\"NOT NULL\")\n if dflt_value is not None:\n parts.append(f\"DEFAULT {dflt_value}\")\n col_defs.append(\" \".join(parts))\n\n cols_quoted = \", \".join(f'\"{c}\"' for c in col_names)\n new_table = f\"{table_name}__new\"\n\n # Create new table and bulk-copy data\n cursor.execute(f'CREATE TABLE \"{new_table}\" ({\", \".join(col_defs)})')\n cursor.execute(\n f'INSERT INTO \"{new_table}\" ({cols_quoted}) '\n f'SELECT {cols_quoted} FROM \"{table_name}\"'\n )\n\n # Translate FK values in the new table\n for old_hash, new_uuid in mapping.items():\n cursor.execute(\n f'UPDATE \"{new_table}\" SET \"{fk_column}\" = ? '\n f'WHERE \"{fk_column}\" = ?',\n [new_uuid, old_hash],\n )\n\n # Swap tables\n cursor.execute(f'DROP TABLE \"{table_name}\"')\n cursor.execute(f'ALTER TABLE \"{new_table}\" RENAME TO \"{table_name}\"')\n\n # Re-create indexes\n for _idx_name, idx_sql in indexes:\n try:\n cursor.execute(idx_sql)\n except Exception:\n pass\n\n\ndef _sqlite_create_indexes(cursor):\n \"\"\"Create the same performance indexes as the PostgreSQL path.\"\"\"\n index_defs = [\n (True, \"api_photo_image_hash_unique\", \"api_photo\", \"image_hash\"),\n (False, \"api_face_photo_id_idx\", \"api_face\", \"photo_id\"),\n (False, \"api_photo_shared_to_photo_id_idx\", \"api_photo_shared_to\", \"photo_id\"),\n (False, \"api_photo_files_photo_id_idx\", \"api_photo_files\", \"photo_id\"),\n (False, \"api_person_cover_photo_id_idx\", \"api_person\", \"cover_photo_id\"),\n (False, \"api_albumuser_cover_photo_id_idx\", \"api_albumuser\", \"cover_photo_id\"),\n (False, \"api_photostack_primary_photo_id_idx\", \"api_photostack\", \"primary_photo_id\"),\n ]\n for unique, idx_name, table, column in index_defs:\n unique_kw = \"UNIQUE \" if unique else \"\"\n try:\n cursor.execute(\n f'CREATE {unique_kw}INDEX IF NOT EXISTS '\n f'\"{idx_name}\" ON \"{table}\"(\"{column}\")'\n )\n except Exception:\n pass\n\n\ndef _sqlite_column_info(cursor, table_name):\n \"\"\"Return PRAGMA table_info rows for *table_name*.\"\"\"\n cursor.execute(f'PRAGMA table_info(\"{table_name}\")')\n return cursor.fetchall()\n\n\ndef _sqlite_index_info(cursor, table_name):\n \"\"\"Return (name, sql) for every explicit index on *table_name*.\"\"\"\n cursor.execute(\n \"SELECT name, sql FROM sqlite_master \"\n \"WHERE type='index' AND tbl_name=? AND sql IS NOT NULL\",\n [table_name],\n )\n return cursor.fetchall()\n\n\n# ============================================================================\n# Migration class\n# ============================================================================\n\nclass Migration(migrations.Migration):\n \"\"\"\n Migration to change Photo primary key from image_hash (CharField) to id (UUIDField).\n\n WARNING: This migration is NOT reversible through Django's migration system.\n You MUST have a database backup before running this migration.\n\n Steps:\n 1. Add UUID column to api_photo\n 2. Generate UUIDs for existing photos\n 3. Add UUID columns to all related tables (to store new FK values)\n 4. Populate new UUID FK columns from image_hash lookups\n 5. Drop all old FK constraints\n 6. Drop old PK, add new PK\n 7. Drop old FK columns, rename new FK columns\n 8. Recreate all FK constraints\n \"\"\"\n\n dependencies = [\n ('api', '0098_add_photo_stack'),\n ]\n\n operations = [\n migrations.SeparateDatabaseAndState(\n database_operations=[\n migrations.RunPython(migrate_forward, migrate_reverse),\n ],\n state_operations=[\n # Add the new UUID primary key field\n migrations.AddField(\n model_name='photo',\n name='id',\n field=models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False),\n ),\n # Change image_hash from primary_key=True to just unique=True\n migrations.AlterField(\n model_name='photo',\n name='image_hash',\n field=models.CharField(db_index=True, max_length=64, unique=True),\n ),\n ],\n ),\n ]\n" }, { "path": "api/migrations/0100_metadataedit_metadatafile_photometadata_stackreview_and_more.py", "content": "# Generated by Django 5.2.9 on 2025-12-25 14:15\n\nimport api.models.user\nimport django.db.models.deletion\nimport uuid\nfrom django.conf import settings\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0099_photo_uuid_primary_key'),\n ]\n\n operations = [\n migrations.CreateModel(\n name='MetadataEdit',\n fields=[\n ('id', models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False)),\n ('field_name', models.CharField(max_length=100)),\n ('old_value', models.JSONField(blank=True, null=True)),\n ('new_value', models.JSONField(blank=True, null=True)),\n ('synced_to_file', models.BooleanField(default=False)),\n ('synced_at', models.DateTimeField(blank=True, null=True)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ],\n options={\n 'verbose_name': 'Metadata Edit',\n 'verbose_name_plural': 'Metadata Edits',\n 'ordering': ['-created_at'],\n },\n ),\n migrations.CreateModel(\n name='MetadataFile',\n fields=[\n ('id', models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False)),\n ('file_type', models.CharField(choices=[('xmp', 'XMP Sidecar'), ('json', 'JSON Metadata'), ('exif', 'EXIF Extract'), ('other', 'Other')], default='xmp', max_length=10)),\n ('source', models.CharField(choices=[('original', 'Original Sidecar'), ('software', 'Software Generated'), ('librephotos', 'LibrePhotos Generated'), ('user', 'User Created')], default='original', max_length=20)),\n ('priority', models.IntegerField(default=0)),\n ('creator_software', models.CharField(blank=True, max_length=100, null=True)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ],\n options={\n 'verbose_name': 'Metadata File',\n 'verbose_name_plural': 'Metadata Files',\n 'ordering': ['-priority', '-updated_at'],\n },\n ),\n migrations.CreateModel(\n name='PhotoMetadata',\n fields=[\n ('id', models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False)),\n ('aperture', models.FloatField(blank=True, db_index=True, null=True)),\n ('shutter_speed', models.CharField(blank=True, max_length=20, null=True)),\n ('shutter_speed_seconds', models.FloatField(blank=True, null=True)),\n ('iso', models.IntegerField(blank=True, db_index=True, null=True)),\n ('focal_length', models.FloatField(blank=True, null=True)),\n ('focal_length_35mm', models.IntegerField(blank=True, null=True)),\n ('exposure_compensation', models.FloatField(blank=True, null=True)),\n ('flash_fired', models.BooleanField(blank=True, null=True)),\n ('metering_mode', models.CharField(blank=True, max_length=50, null=True)),\n ('white_balance', models.CharField(blank=True, max_length=50, null=True)),\n ('camera_make', models.CharField(blank=True, db_index=True, max_length=100, null=True)),\n ('camera_model', models.CharField(blank=True, db_index=True, max_length=100, null=True)),\n ('lens_make', models.CharField(blank=True, max_length=100, null=True)),\n ('lens_model', models.CharField(blank=True, db_index=True, max_length=100, null=True)),\n ('serial_number', models.CharField(blank=True, max_length=100, null=True)),\n ('width', models.IntegerField(blank=True, null=True)),\n ('height', models.IntegerField(blank=True, null=True)),\n ('orientation', models.IntegerField(blank=True, null=True)),\n ('color_space', models.CharField(blank=True, max_length=50, null=True)),\n ('bit_depth', models.IntegerField(blank=True, null=True)),\n ('date_taken', models.DateTimeField(blank=True, db_index=True, null=True)),\n ('date_taken_subsec', models.CharField(blank=True, max_length=10, null=True)),\n ('date_modified', models.DateTimeField(blank=True, null=True)),\n ('timezone_offset', models.CharField(blank=True, max_length=10, null=True)),\n ('gps_latitude', models.FloatField(blank=True, db_index=True, null=True)),\n ('gps_longitude', models.FloatField(blank=True, db_index=True, null=True)),\n ('gps_altitude', models.FloatField(blank=True, null=True)),\n ('location_country', models.CharField(blank=True, db_index=True, max_length=100, null=True)),\n ('location_state', models.CharField(blank=True, max_length=100, null=True)),\n ('location_city', models.CharField(blank=True, db_index=True, max_length=100, null=True)),\n ('location_address', models.TextField(blank=True, null=True)),\n ('title', models.CharField(blank=True, max_length=500, null=True)),\n ('caption', models.TextField(blank=True, null=True)),\n ('keywords', models.JSONField(blank=True, null=True)),\n ('rating', models.IntegerField(blank=True, db_index=True, null=True)),\n ('copyright', models.TextField(blank=True, null=True)),\n ('creator', models.CharField(blank=True, max_length=200, null=True)),\n ('source', models.CharField(choices=[('embedded', 'Embedded in File'), ('sidecar', 'XMP Sidecar'), ('user_edit', 'User Edit'), ('computed', 'Computed')], default='embedded', max_length=20)),\n ('raw_exif', models.JSONField(blank=True, null=True)),\n ('raw_xmp', models.JSONField(blank=True, null=True)),\n ('raw_iptc', models.JSONField(blank=True, null=True)),\n ('version', models.IntegerField(default=1)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ],\n options={\n 'verbose_name': 'Photo Metadata',\n 'verbose_name_plural': 'Photo Metadata',\n },\n ),\n migrations.CreateModel(\n name='StackReview',\n fields=[\n ('id', models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False)),\n ('decision', models.CharField(choices=[('pending', 'Pending Review'), ('resolved', 'Resolved'), ('dismissed', 'Dismissed')], db_index=True, default='pending', max_length=20)),\n ('trashed_count', models.IntegerField(default=0)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('reviewed_at', models.DateTimeField(blank=True, null=True)),\n ('note', models.TextField(blank=True, null=True)),\n ],\n options={\n 'verbose_name': 'Stack Review',\n 'verbose_name_plural': 'Stack Reviews',\n 'ordering': ['-created_at'],\n },\n ),\n migrations.RemoveIndex(\n model_name='photostack',\n name='api_photost_owner_i_abc123_idx',\n ),\n migrations.RemoveIndex(\n model_name='photostack',\n name='api_photost_owner_i_def456_idx',\n ),\n migrations.RemoveField(\n model_name='photostack',\n name='status',\n ),\n migrations.AlterField(\n model_name='photostack',\n name='owner',\n field=models.ForeignKey(on_delete=models.SET(api.models.user.get_deleted_user), related_name='photo_stacks', to=settings.AUTH_USER_MODEL),\n ),\n migrations.AddIndex(\n model_name='photostack',\n index=models.Index(fields=['owner', 'stack_type'], name='api_photost_owner_i_40a369_idx'),\n ),\n migrations.AddField(\n model_name='metadataedit',\n name='photo',\n field=models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='metadata_edits', to='api.photo'),\n ),\n migrations.AddField(\n model_name='metadataedit',\n name='user',\n field=models.ForeignKey(on_delete=models.SET(api.models.user.get_deleted_user), related_name='metadata_edits', to=settings.AUTH_USER_MODEL),\n ),\n migrations.AddField(\n model_name='metadatafile',\n name='file',\n field=models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, related_name='metadata_info', to='api.file'),\n ),\n migrations.AddField(\n model_name='metadatafile',\n name='photo',\n field=models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='metadata_files', to='api.photo'),\n ),\n migrations.AddField(\n model_name='photometadata',\n name='photo',\n field=models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, related_name='metadata', to='api.photo'),\n ),\n migrations.AddField(\n model_name='stackreview',\n name='kept_photo',\n field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='kept_in_reviews', to='api.photo'),\n ),\n migrations.AddField(\n model_name='stackreview',\n name='reviewer',\n field=models.ForeignKey(on_delete=models.SET(api.models.user.get_deleted_user), related_name='stack_reviews', to=settings.AUTH_USER_MODEL),\n ),\n migrations.AddField(\n model_name='stackreview',\n name='stack',\n field=models.OneToOneField(on_delete=django.db.models.deletion.CASCADE, related_name='review', to='api.photostack'),\n ),\n migrations.AddIndex(\n model_name='metadataedit',\n index=models.Index(fields=['photo', '-created_at'], name='api_metadat_photo_i_21e478_idx'),\n ),\n migrations.AddIndex(\n model_name='photometadata',\n index=models.Index(fields=['camera_make', 'camera_model'], name='api_photome_camera__361eac_idx'),\n ),\n migrations.AddIndex(\n model_name='photometadata',\n index=models.Index(fields=['date_taken'], name='api_photome_date_ta_35e59b_idx'),\n ),\n migrations.AddIndex(\n model_name='photometadata',\n index=models.Index(fields=['location_country', 'location_city'], name='api_photome_locatio_1ca8bb_idx'),\n ),\n migrations.AddIndex(\n model_name='stackreview',\n index=models.Index(fields=['reviewer', 'decision'], name='api_stackre_reviewe_48380f_idx'),\n ),\n ]\n" }, { "path": "api/migrations/0101_populate_photo_metadata.py", "content": "# Generated migration to populate PhotoMetadata from existing Photo data\n\nfrom django.db import migrations, transaction\nfrom django.db.models import Exists, OuterRef\n\n\nBATCH_SIZE = 1000\n\n\ndef populate_photo_metadata(apps, schema_editor):\n \"\"\"\n Populate PhotoMetadata for all existing photos.\n\n This copies metadata fields from Photo model to the structured PhotoMetadata model.\n PhotoMetadata provides:\n - Normalized field names\n - Edit history tracking\n - XMP sidecar support\n - Better organization of camera/lens/settings\n\n Optimized for SQLite and PostgreSQL compatibility:\n - Fetches all photo IDs upfront (avoids open cursor + writes conflict on SQLite)\n - Loads caption data per-batch to avoid memory issues\n - Processes in batches with per-batch transactions (no huge outer transaction)\n \"\"\"\n Photo = apps.get_model(\"api\", \"Photo\")\n PhotoMetadata = apps.get_model(\"api\", \"PhotoMetadata\")\n PhotoCaption = apps.get_model(\"api\", \"PhotoCaption\")\n\n # Exists subquery for efficient filtering\n existing_metadata = PhotoMetadata.objects.filter(photo_id=OuterRef('pk'))\n\n # Collect all photo IDs that need metadata upfront.\n # Using values_list avoids keeping a cursor open while we write below,\n # which can cause SQLite \"database is locked\" / cursor-state issues.\n photo_ids = list(\n Photo.objects\n .filter(~Exists(existing_metadata))\n .values_list('pk', flat=True)\n )\n\n total_count = len(photo_ids)\n if total_count == 0:\n print(\"No photos need metadata population.\")\n return\n\n print(f\"Populating metadata for {total_count} photos...\")\n\n processed = 0\n\n for i in range(0, total_count, BATCH_SIZE):\n chunk_ids = photo_ids[i:i + BATCH_SIZE]\n\n # Load caption data for this chunk only\n captions = {\n c.photo_id: c.captions_json\n for c in PhotoCaption.objects.filter(photo_id__in=chunk_ids)\n }\n\n batch = []\n for photo in Photo.objects.filter(pk__in=chunk_ids):\n captions_json = captions.get(photo.pk)\n metadata = PhotoMetadata(\n photo=photo,\n # Camera info\n camera_make=None, # Not stored in Photo model separately\n camera_model=photo.camera,\n lens_make=None, # Not stored separately\n lens_model=photo.lens,\n # Capture settings\n aperture=photo.fstop,\n shutter_speed=photo.shutter_speed,\n iso=photo.iso,\n focal_length=photo.focal_length,\n focal_length_35mm=photo.focalLength35Equivalent,\n # Image properties\n width=photo.width,\n height=photo.height,\n # Date/time\n date_taken=photo.exif_timestamp,\n # GPS\n gps_latitude=photo.exif_gps_lat,\n gps_longitude=photo.exif_gps_lon,\n # Content\n title=None, # Photo doesn't have separate title\n caption=captions_json.get(\"user_caption\") if captions_json else None,\n keywords=list(captions_json.get(\"keywords\", [])) if captions_json else [],\n rating=photo.rating,\n # Source\n source=\"embedded\", # All existing data came from EXIF\n version=1,\n )\n batch.append(metadata)\n\n with transaction.atomic():\n PhotoMetadata.objects.bulk_create(batch, ignore_conflicts=True)\n processed += len(batch)\n print(f\" Processed {processed}/{total_count} photos ({100*processed//total_count}%)\")\n\n print(f\"Completed populating metadata for {processed} photos.\")\n\n\ndef reverse_populate(apps, schema_editor):\n \"\"\"\n Reverse migration - delete PhotoMetadata records.\n Note: This will lose any user edits made through PhotoMetadata.\n \"\"\"\n PhotoMetadata = apps.get_model(\"api\", \"PhotoMetadata\")\n PhotoMetadata.objects.all().delete()\n\n\nclass Migration(migrations.Migration):\n \"\"\"\n Data migration to populate PhotoMetadata from existing Photo data.\n \n This ensures backwards compatibility:\n - Photo model still has all the original fields\n - PhotoMetadata provides structured access + edit history\n - API can read from either, preferring PhotoMetadata when available\n \"\"\"\n\n dependencies = [\n (\"api\", \"0100_metadataedit_metadatafile_photometadata_stackreview_and_more\"),\n ]\n\n operations = [\n migrations.RunPython(\n populate_photo_metadata,\n reverse_populate,\n atomic=False,\n ),\n ]\n" }, { "path": "api/migrations/0102_photo_stacks_manytomany.py", "content": "\"\"\"\nMigration to convert Photo.stack ForeignKey to Photo.stacks ManyToManyField.\n\nThis change allows a photo to belong to multiple stacks of different types\nsimultaneously, preventing data loss when photos have multiple relationships:\n- A RAW+JPEG pair can also be visually similar to other photos\n- A burst sequence can also have exact copies\n- etc.\n\"\"\"\n\nfrom django.db import migrations, models\n\n\ndef migrate_fk_to_m2m(apps, schema_editor):\n \"\"\"\n Migrate existing ForeignKey relationships to ManyToMany.\n \n For each photo that has a stack ForeignKey set, add that stack\n to the new ManyToMany relationship.\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n \n # Get all photos with a stack set (using the old FK field)\n photos_with_stacks = Photo.objects.filter(stack__isnull=False).select_related('stack')\n \n for photo in photos_with_stacks:\n # Add the old FK stack to the new M2M relationship\n photo.stacks.add(photo.stack)\n\n\ndef reverse_m2m_to_fk(apps, schema_editor):\n \"\"\"\n Reverse migration: convert ManyToMany back to ForeignKey.\n \n For each photo, set the FK to the first stack in the M2M relationship.\n Note: This may lose data if a photo was in multiple stacks.\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n \n for photo in Photo.objects.prefetch_related('stacks').all():\n first_stack = photo.stacks.first()\n if first_stack:\n photo.stack = first_stack\n photo.save(update_fields=['stack'])\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0101_populate_photo_metadata'),\n ]\n\n operations = [\n # Step 1: Add the new ManyToMany field\n migrations.AddField(\n model_name='photo',\n name='stacks',\n field=models.ManyToManyField(\n blank=True,\n related_name='photos_m2m',\n to='api.photostack',\n ),\n ),\n \n # Step 2: Migrate data from FK to M2M\n migrations.RunPython(\n migrate_fk_to_m2m,\n reverse_m2m_to_fk,\n ),\n \n # Step 3: Remove the old ForeignKey field\n migrations.RemoveField(\n model_name='photo',\n name='stack',\n ),\n \n # Step 4: Rename M2M related_name to final name\n migrations.AlterField(\n model_name='photo',\n name='stacks',\n field=models.ManyToManyField(\n blank=True,\n related_name='photos',\n to='api.photostack',\n ),\n ),\n ]\n" }, { "path": "api/migrations/0103_remove_photo_metadata_fields.py", "content": "# Generated migration to remove deprecated metadata fields from Photo model\n# These fields have been migrated to PhotoMetadata model\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n \"\"\"\n Remove deprecated metadata fields from Photo model.\n \n These fields have been migrated to the structured PhotoMetadata model:\n - camera -> PhotoMetadata.camera_model\n - lens -> PhotoMetadata.lens_model\n - fstop -> PhotoMetadata.aperture\n - shutter_speed -> PhotoMetadata.shutter_speed\n - iso -> PhotoMetadata.iso\n - focal_length -> PhotoMetadata.focal_length\n - focalLength35Equivalent -> PhotoMetadata.focal_length_35mm\n - width -> PhotoMetadata.width\n - height -> PhotoMetadata.height\n - digitalZoomRatio -> Not migrated (rarely used)\n - subjectDistance -> Not migrated (rarely used)\n \n Data was already copied in migration 0101_populate_photo_metadata.\n \"\"\"\n\n dependencies = [\n (\"api\", \"0102_photo_stacks_manytomany\"),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"fstop\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"focal_length\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"iso\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"shutter_speed\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"camera\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"lens\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"width\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"height\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"focalLength35Equivalent\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"digitalZoomRatio\",\n ),\n migrations.RemoveField(\n model_name=\"photo\",\n name=\"subjectDistance\",\n ),\n ]\n" }, { "path": "api/migrations/0104_remove_photostack_potential_savings_and_more.py", "content": "# Generated by Django 5.2.9 on 2025-12-26 15:05\n\nimport api.models.user\nimport django.db.models.deletion\nimport uuid\nfrom django.conf import settings\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0103_remove_photo_metadata_fields'),\n ]\n\n operations = [\n migrations.RemoveField(\n model_name='photostack',\n name='potential_savings',\n ),\n migrations.RemoveField(\n model_name='photostack',\n name='similarity_score',\n ),\n migrations.AlterField(\n model_name='photostack',\n name='stack_type',\n field=models.CharField(choices=[('raw_jpeg', 'RAW + JPEG Pair'), ('burst', 'Burst Sequence'), ('bracket', 'Exposure Bracket'), ('live_photo', 'Live Photo'), ('manual', 'Manual Stack')], db_index=True, default='raw_jpeg', max_length=20),\n ),\n migrations.CreateModel(\n name='Duplicate',\n fields=[\n ('id', models.UUIDField(default=uuid.uuid4, editable=False, primary_key=True, serialize=False)),\n ('duplicate_type', models.CharField(choices=[('exact_copy', 'Exact Copies'), ('visual_duplicate', 'Visual Duplicates')], db_index=True, default='visual_duplicate', max_length=20)),\n ('review_status', models.CharField(choices=[('pending', 'Pending Review'), ('resolved', 'Resolved'), ('dismissed', 'Dismissed')], db_index=True, default='pending', max_length=20)),\n ('created_at', models.DateTimeField(auto_now_add=True)),\n ('updated_at', models.DateTimeField(auto_now=True)),\n ('reviewed_at', models.DateTimeField(blank=True, null=True)),\n ('similarity_score', models.FloatField(blank=True, null=True)),\n ('potential_savings', models.BigIntegerField(default=0)),\n ('trashed_count', models.IntegerField(default=0)),\n ('note', models.TextField(blank=True, null=True)),\n ('kept_photo', models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='kept_in_duplicates', to='api.photo')),\n ('owner', models.ForeignKey(on_delete=models.SET(api.models.user.get_deleted_user), related_name='duplicates', to=settings.AUTH_USER_MODEL)),\n ],\n options={\n 'verbose_name': 'Duplicate',\n 'verbose_name_plural': 'Duplicates',\n 'ordering': ['-created_at'],\n },\n ),\n migrations.AddField(\n model_name='photo',\n name='duplicates',\n field=models.ManyToManyField(blank=True, related_name='photos', to='api.duplicate'),\n ),\n migrations.AddIndex(\n model_name='duplicate',\n index=models.Index(fields=['owner', 'duplicate_type'], name='api_duplica_owner_i_78a3a4_idx'),\n ),\n migrations.AddIndex(\n model_name='duplicate',\n index=models.Index(fields=['owner', 'review_status'], name='api_duplica_owner_i_039fa3_idx'),\n ),\n ]\n" }, { "path": "api/migrations/0105_alter_photo_image_hash.py", "content": "# Generated by Django 5.2.9 on 2025-12-26 16:12\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0104_remove_photostack_potential_savings_and_more'),\n ]\n\n operations = [\n # Drop the unique index created by migration 0099 (it was created as an index, not a constraint)\n migrations.RunSQL(\n sql=\"DROP INDEX IF EXISTS api_photo_image_hash_unique;\",\n reverse_sql=\"CREATE UNIQUE INDEX IF NOT EXISTS api_photo_image_hash_unique ON api_photo(image_hash);\",\n ),\n migrations.AlterField(\n model_name='photo',\n name='image_hash',\n field=models.CharField(db_index=True, max_length=64),\n ),\n ]\n" }, { "path": "api/migrations/0106_alter_longrunningjob_options.py", "content": "# Generated by Django 5.2.9 on 2025-12-26 19:35\n\nfrom django.db import migrations\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0105_alter_photo_image_hash'),\n ]\n\n operations = [\n migrations.AlterModelOptions(\n name='longrunningjob',\n options={'ordering': ['-queued_at'], 'verbose_name': 'Long Running Job', 'verbose_name_plural': 'Long Running Jobs'},\n ),\n ]\n" }, { "path": "api/migrations/0107_add_burst_detection_rules.py", "content": "# Generated by Django 5.0 on 2024-12-26\n\nfrom django.db import migrations, models\n\nimport api.models.user\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0106_alter_longrunningjob_options\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"burst_detection_rules\",\n field=models.JSONField(\n default=api.models.user.get_default_config_burst_detection_rules\n ),\n ),\n ]\n" }, { "path": "api/migrations/0108_add_stack_raw_jpeg_field.py", "content": "# Generated migration for adding stack_raw_jpeg field\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0107_add_burst_detection_rules\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"stack_raw_jpeg\",\n field=models.BooleanField(default=True),\n ),\n ]\n\n\n" }, { "path": "api/migrations/0109_migrate_skip_raw_to_stack_raw_jpeg.py", "content": "# Data migration to set stack_raw_jpeg based on skip_raw_files\n# If skip_raw_files was True (skip RAWs), then stack_raw_jpeg should be False (don't stack)\n# If skip_raw_files was False (don't skip RAWs), then stack_raw_jpeg should be True (stack them)\n\nfrom django.db import migrations\n\n\ndef migrate_skip_raw_to_stack_raw_jpeg(apps, schema_editor):\n User = apps.get_model(\"api\", \"User\")\n # Set stack_raw_jpeg = not skip_raw_files\n # If user was skipping RAW files, they probably don't want them stacked\n # If user was not skipping RAW files, enable stacking by default\n User.objects.filter(skip_raw_files=True).update(stack_raw_jpeg=False)\n User.objects.filter(skip_raw_files=False).update(stack_raw_jpeg=True)\n\n\ndef reverse_migration(apps, schema_editor):\n # Reverse migration: set skip_raw_files based on stack_raw_jpeg\n User = apps.get_model(\"api\", \"User\")\n User.objects.filter(stack_raw_jpeg=False).update(skip_raw_files=True)\n User.objects.filter(stack_raw_jpeg=True).update(skip_raw_files=False)\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0108_add_stack_raw_jpeg_field\"),\n ]\n\n operations = [\n migrations.RunPython(migrate_skip_raw_to_stack_raw_jpeg, reverse_migration),\n ]\n\n\n" }, { "path": "api/migrations/0110_fix_file_embedded_media_self_reference.py", "content": "# Generated migration to fix self-referential ManyToManyField\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n dependencies = [\n (\"api\", \"0109_migrate_skip_raw_to_stack_raw_jpeg\"),\n ]\n\n operations = [\n migrations.AlterField(\n model_name=\"file\",\n name=\"embedded_media\",\n field=models.ManyToManyField(\"self\", symmetrical=False),\n ),\n ]\n" }, { "path": "api/migrations/0111_alter_file_embedded_media.py", "content": "# Generated by Django 5.2.9 on 2026-01-08 19:16\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0110_fix_file_embedded_media_self_reference'),\n ]\n\n operations = [\n migrations.AlterField(\n model_name='file',\n name='embedded_media',\n field=models.ManyToManyField(to='api.file'),\n ),\n ]\n" }, { "path": "api/migrations/0112_convert_file_stacks_to_variants.py", "content": "\"\"\"\nMigration to convert RAW_JPEG_PAIR and LIVE_PHOTO stacks to file variants.\n\nThis migration implements the PhotoPrism-like file variant model where:\n- RAW+JPEG pairs become one Photo with multiple files\n- Live Photos (image+video) become one Photo with multiple files\n\nInstead of having 2 Photo entities in a stack, we now have 1 Photo entity\nwith multiple File entries in its files ManyToMany field.\n\nThis is a data migration that:\n1. For each RAW_JPEG_PAIR stack:\n - Identifies the JPEG Photo (primary) and RAW Photo\n - Moves the RAW file to the JPEG Photo's files field\n - Deletes the RAW Photo entity\n - Deletes the stack\n2. For each LIVE_PHOTO stack:\n - Identifies the image Photo (primary) and video Photo\n - Moves the video file to the image Photo's files field\n - Deletes the video Photo entity\n - Deletes the stack\n\nOptimized for large datasets:\n- Uses prefetch_related to eliminate N+1 queries\n- Uses bulk M2M operations via through model\n- Processes in batches with progress logging\n\"\"\"\n\nfrom django.db import migrations, transaction\nfrom django.db.models import Prefetch\n\n\nBATCH_SIZE = 500\n\n\ndef convert_raw_jpeg_stacks_to_file_variants(apps, schema_editor):\n \"\"\"Convert RAW_JPEG_PAIR stacks to Photo.files variants.\"\"\"\n PhotoStack = apps.get_model('api', 'PhotoStack')\n Photo = apps.get_model('api', 'Photo')\n File = apps.get_model('api', 'File')\n \n # Get through model for bulk M2M operations\n PhotoFiles = Photo.files.through\n PhotoStacks = Photo.stacks.through\n \n # RAW_JPEG_PAIR = \"raw_jpeg\"\n # Count total for progress logging\n total_count = PhotoStack.objects.filter(stack_type=\"raw_jpeg\").count()\n if total_count == 0:\n print(\"No RAW_JPEG_PAIR stacks to convert.\")\n return\n \n print(f\"Converting {total_count} RAW_JPEG_PAIR stacks to file variants...\")\n \n # Prefetch photos with their files and main_file to eliminate N+1 queries\n raw_jpeg_stacks = (\n PhotoStack.objects\n .filter(stack_type=\"raw_jpeg\")\n .prefetch_related(\n Prefetch(\n 'photos',\n queryset=Photo.objects.select_related('main_file').prefetch_related('files')\n )\n )\n )\n \n converted_count = 0\n error_count = 0\n \n # Collect bulk operations\n m2m_files_to_create = []\n photos_to_delete = []\n stacks_to_delete = []\n m2m_stacks_to_delete = []\n \n for stack in raw_jpeg_stacks.iterator(chunk_size=BATCH_SIZE):\n try:\n # Photos are already prefetched - no extra query\n photos = list(stack.photos.all())\n \n if len(photos) != 2:\n print(f\"WARNING: RAW_JPEG stack {stack.id} has {len(photos)} photos, expected 2. Skipping.\")\n error_count += 1\n continue\n \n # Identify JPEG and RAW photos\n # RAW files have type=4 in File model\n jpeg_photo = None\n raw_photo = None\n \n for photo in photos:\n if photo.main_file and photo.main_file.type == 4: # RAW_FILE\n raw_photo = photo\n else:\n jpeg_photo = photo\n \n if not jpeg_photo or not raw_photo:\n print(f\"WARNING: Could not identify JPEG/RAW in stack {stack.id}. Skipping.\")\n error_count += 1\n continue\n \n # Collect files to add to jpeg_photo (using prefetched data)\n files_to_add = list(raw_photo.files.all())\n if raw_photo.main_file:\n files_to_add.append(raw_photo.main_file)\n \n # Build M2M through model entries for bulk create\n existing_file_hashes = set(f.hash for f in jpeg_photo.files.all())\n for file in files_to_add:\n if file.hash not in existing_file_hashes:\n m2m_files_to_create.append(\n PhotoFiles(photo_id=jpeg_photo.pk, file_id=file.hash)\n )\n existing_file_hashes.add(file.hash)\n \n # Collect M2M stack relationships to delete\n for photo in photos:\n m2m_stacks_to_delete.append((photo.pk, stack.pk))\n \n photos_to_delete.append(raw_photo.pk)\n stacks_to_delete.append(stack.pk)\n converted_count += 1\n \n # Process in batches to avoid memory buildup\n if len(stacks_to_delete) >= BATCH_SIZE:\n _flush_raw_jpeg_batch(\n PhotoFiles, PhotoStacks, Photo, PhotoStack,\n m2m_files_to_create, m2m_stacks_to_delete, photos_to_delete, stacks_to_delete\n )\n m2m_files_to_create = []\n m2m_stacks_to_delete = []\n photos_to_delete = []\n stacks_to_delete = []\n print(f\" Processed {converted_count}/{total_count} stacks ({100*converted_count//total_count}%)\")\n \n except Exception as e:\n print(f\"ERROR converting RAW_JPEG stack {stack.id}: {e}\")\n error_count += 1\n \n # Flush remaining batch\n if stacks_to_delete:\n _flush_raw_jpeg_batch(\n PhotoFiles, PhotoStacks, Photo, PhotoStack,\n m2m_files_to_create, m2m_stacks_to_delete, photos_to_delete, stacks_to_delete\n )\n \n print(f\"Converted {converted_count} RAW_JPEG_PAIR stacks to file variants ({error_count} errors)\")\n\n\ndef _flush_raw_jpeg_batch(PhotoFiles, PhotoStacks, Photo, PhotoStack,\n m2m_files_to_create, m2m_stacks_to_delete, photos_to_delete, stacks_to_delete):\n \"\"\"Flush a batch of operations to the database.\"\"\"\n with transaction.atomic():\n # Bulk create M2M file relationships\n if m2m_files_to_create:\n PhotoFiles.objects.bulk_create(m2m_files_to_create, ignore_conflicts=True)\n \n # Bulk delete M2M stack relationships\n if m2m_stacks_to_delete:\n for photo_id, stack_id in m2m_stacks_to_delete:\n PhotoStacks.objects.filter(photo_id=photo_id, photostack_id=stack_id).delete()\n \n # Bulk delete photos\n if photos_to_delete:\n Photo.objects.filter(pk__in=photos_to_delete).delete()\n \n # Bulk delete stacks\n if stacks_to_delete:\n PhotoStack.objects.filter(pk__in=stacks_to_delete).delete()\n\n\ndef convert_live_photo_stacks_to_file_variants(apps, schema_editor):\n \"\"\"Convert LIVE_PHOTO stacks to Photo.files variants.\"\"\"\n PhotoStack = apps.get_model('api', 'PhotoStack')\n Photo = apps.get_model('api', 'Photo')\n File = apps.get_model('api', 'File')\n \n # Get through model for bulk M2M operations\n PhotoFiles = Photo.files.through\n PhotoStacks = Photo.stacks.through\n \n # LIVE_PHOTO = \"live_photo\"\n # Count total for progress logging\n total_count = PhotoStack.objects.filter(stack_type=\"live_photo\").count()\n if total_count == 0:\n print(\"No LIVE_PHOTO stacks to convert.\")\n return\n \n print(f\"Converting {total_count} LIVE_PHOTO stacks to file variants...\")\n \n # Prefetch photos with their files and main_file to eliminate N+1 queries\n live_photo_stacks = (\n PhotoStack.objects\n .filter(stack_type=\"live_photo\")\n .prefetch_related(\n Prefetch(\n 'photos',\n queryset=Photo.objects.select_related('main_file').prefetch_related('files')\n )\n )\n )\n \n converted_count = 0\n error_count = 0\n \n # Collect bulk operations\n m2m_files_to_create = []\n photos_to_delete = []\n stacks_to_delete = []\n m2m_stacks_to_delete = []\n \n for stack in live_photo_stacks.iterator(chunk_size=BATCH_SIZE):\n try:\n # Photos are already prefetched - no extra query\n photos = list(stack.photos.all())\n \n if len(photos) != 2:\n print(f\"WARNING: LIVE_PHOTO stack {stack.id} has {len(photos)} photos, expected 2. Skipping.\")\n error_count += 1\n continue\n \n # Identify image and video photos\n # VIDEO files have type=2 in File model\n image_photo = None\n video_photo = None\n \n for photo in photos:\n if photo.main_file and photo.main_file.type == 2: # VIDEO\n video_photo = photo\n elif photo.video:\n video_photo = photo\n else:\n image_photo = photo\n \n if not image_photo or not video_photo:\n print(f\"WARNING: Could not identify image/video in LIVE_PHOTO stack {stack.id}. Skipping.\")\n error_count += 1\n continue\n \n # Collect files to add to image_photo (using prefetched data)\n files_to_add = list(video_photo.files.all())\n if video_photo.main_file:\n files_to_add.append(video_photo.main_file)\n \n # Build M2M through model entries for bulk create\n existing_file_hashes = set(f.hash for f in image_photo.files.all())\n for file in files_to_add:\n if file.hash not in existing_file_hashes:\n m2m_files_to_create.append(\n PhotoFiles(photo_id=image_photo.pk, file_id=file.hash)\n )\n existing_file_hashes.add(file.hash)\n \n # Collect M2M stack relationships to delete\n for photo in photos:\n m2m_stacks_to_delete.append((photo.pk, stack.pk))\n \n photos_to_delete.append(video_photo.pk)\n stacks_to_delete.append(stack.pk)\n converted_count += 1\n \n # Process in batches to avoid memory buildup\n if len(stacks_to_delete) >= BATCH_SIZE:\n _flush_live_photo_batch(\n PhotoFiles, PhotoStacks, Photo, PhotoStack,\n m2m_files_to_create, m2m_stacks_to_delete, photos_to_delete, stacks_to_delete\n )\n m2m_files_to_create = []\n m2m_stacks_to_delete = []\n photos_to_delete = []\n stacks_to_delete = []\n print(f\" Processed {converted_count}/{total_count} stacks ({100*converted_count//total_count}%)\")\n \n except Exception as e:\n print(f\"ERROR converting LIVE_PHOTO stack {stack.id}: {e}\")\n error_count += 1\n \n # Flush remaining batch\n if stacks_to_delete:\n _flush_live_photo_batch(\n PhotoFiles, PhotoStacks, Photo, PhotoStack,\n m2m_files_to_create, m2m_stacks_to_delete, photos_to_delete, stacks_to_delete\n )\n \n print(f\"Converted {converted_count} LIVE_PHOTO stacks to file variants ({error_count} errors)\")\n\n\ndef _flush_live_photo_batch(PhotoFiles, PhotoStacks, Photo, PhotoStack,\n m2m_files_to_create, m2m_stacks_to_delete, photos_to_delete, stacks_to_delete):\n \"\"\"Flush a batch of operations to the database.\"\"\"\n with transaction.atomic():\n # Bulk create M2M file relationships\n if m2m_files_to_create:\n PhotoFiles.objects.bulk_create(m2m_files_to_create, ignore_conflicts=True)\n \n # Bulk delete M2M stack relationships\n if m2m_stacks_to_delete:\n for photo_id, stack_id in m2m_stacks_to_delete:\n PhotoStacks.objects.filter(photo_id=photo_id, photostack_id=stack_id).delete()\n \n # Bulk delete photos\n if photos_to_delete:\n Photo.objects.filter(pk__in=photos_to_delete).delete()\n \n # Bulk delete stacks\n if stacks_to_delete:\n PhotoStack.objects.filter(pk__in=stacks_to_delete).delete()\n\n\ndef forward_migration(apps, schema_editor):\n \"\"\"Run both conversions.\"\"\"\n convert_raw_jpeg_stacks_to_file_variants(apps, schema_editor)\n convert_live_photo_stacks_to_file_variants(apps, schema_editor)\n\n\ndef reverse_migration(apps, schema_editor):\n \"\"\"\n Reverse migration is not fully supported as we've deleted Photo entities.\n This would require recreating the deleted Photos which is complex.\n Instead, we just print a warning.\n \"\"\"\n print(\"WARNING: Reverse migration is not supported. \"\n \"RAW_JPEG_PAIR and LIVE_PHOTO stacks cannot be recreated automatically. \"\n \"Run a full rescan to detect file variants again.\")\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0111_alter_file_embedded_media'),\n ]\n\n operations = [\n migrations.RunPython(forward_migration, reverse_migration),\n ]\n" }, { "path": "api/migrations/0113_alter_photostack_stack_type.py", "content": "# Generated by Django 5.2.9 on 2026-01-21 13:36\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0112_convert_file_stacks_to_variants'),\n ]\n\n operations = [\n migrations.AlterField(\n model_name='photostack',\n name='stack_type',\n field=models.CharField(choices=[('burst', 'Burst Sequence'), ('bracket', 'Exposure Bracket'), ('manual', 'Manual Stack'), ('raw_jpeg', 'RAW + JPEG Pair (Deprecated)'), ('live_photo', 'Live Photo (Deprecated)')], db_index=True, default='manual', max_length=20),\n ),\n ]\n" }, { "path": "api/migrations/0114_add_file_path_unique.py", "content": "# Generated migration to add unique constraint on File.path\n# This migration handles existing duplicate paths before adding the constraint\n\nfrom django.db import migrations, models, transaction\nfrom django.db.models import Count, Exists, OuterRef, Case, When, Value, IntegerField\n\n\nBATCH_SIZE = 500\n\n\ndef deduplicate_file_paths(apps, schema_editor):\n \"\"\"\n Deduplicate File records that have the same path.\n \n Strategy:\n 1. Find all paths that have multiple File records\n 2. For each duplicate group, keep the \"best\" File:\n - Prefer non-missing files\n - Prefer files that are main_file for a Photo\n - Prefer files linked to more Photos\n 3. Reassign all Photo associations from deleted Files to kept File\n 4. Delete duplicate File records\n \n Optimized for large datasets:\n - Uses annotations to compute scores in database instead of per-file queries\n - Uses bulk M2M operations via through model\n - Uses prefetch_related to eliminate N+1 queries\n - Processes in batches with progress logging\n \"\"\"\n File = apps.get_model('api', 'File')\n Photo = apps.get_model('api', 'Photo')\n \n # Get through models for bulk M2M operations\n PhotoFiles = Photo.files.through\n FileEmbeddedMedia = File.embedded_media.through\n \n # Find paths that have duplicates (excluding empty paths)\n duplicate_paths = list(\n File.objects\n .exclude(path='')\n .exclude(path__isnull=True)\n .values('path')\n .annotate(count=Count('hash'))\n .filter(count__gt=1)\n )\n \n total_count = len(duplicate_paths)\n if total_count == 0:\n print(\"No duplicate file paths to deduplicate.\")\n return\n \n print(f\"Deduplicating {total_count} file paths...\")\n \n processed = 0\n deleted_count = 0\n \n for dup in duplicate_paths:\n path = dup['path']\n \n # Subquery to check if file is main_file for any Photo\n is_main_file = Photo.objects.filter(main_file_id=OuterRef('hash'))\n \n # Get all Files with this path, annotated with scores computed in DB\n files = list(\n File.objects\n .filter(path=path)\n .annotate(\n photo_count=Count('photo', distinct=True),\n is_main=Case(\n When(Exists(is_main_file), then=Value(50)),\n default=Value(0),\n output_field=IntegerField()\n ),\n missing_penalty=Case(\n When(missing=False, then=Value(100)),\n default=Value(0),\n output_field=IntegerField()\n ),\n )\n .prefetch_related('photo_set', 'embedded_media')\n .order_by('-missing_penalty', '-is_main', '-photo_count')\n )\n \n if len(files) <= 1:\n continue\n \n # First file is the best one (sorted by score descending)\n keep_file = files[0]\n delete_files = files[1:]\n \n with transaction.atomic():\n # Collect M2M entries to create\n m2m_photo_files_to_create = []\n m2m_photo_files_to_delete = []\n m2m_embedded_to_create = []\n m2m_embedded_to_delete = []\n photos_to_update_main = []\n \n for del_file in delete_files:\n # Get all Photos that have this file in their files M2M (prefetched)\n for photo in del_file.photo_set.all():\n # Schedule add of keep_file to photo\n m2m_photo_files_to_create.append(\n PhotoFiles(photo_id=photo.pk, file_id=keep_file.hash)\n )\n # Schedule removal of del_file from photo\n m2m_photo_files_to_delete.append((photo.pk, del_file.hash))\n \n # Update main_file references in bulk\n photos_with_main = Photo.objects.filter(main_file=del_file)\n for photo in photos_with_main:\n photo.main_file = keep_file\n photos_to_update_main.append(photo)\n \n # Handle embedded_media M2M (prefetched)\n for parent_file in File.objects.filter(embedded_media=del_file):\n m2m_embedded_to_create.append(\n FileEmbeddedMedia(from_file_id=parent_file.hash, to_file_id=keep_file.hash)\n )\n m2m_embedded_to_delete.append((parent_file.hash, del_file.hash))\n \n # Execute bulk operations\n if m2m_photo_files_to_create:\n PhotoFiles.objects.bulk_create(m2m_photo_files_to_create, ignore_conflicts=True)\n \n if m2m_photo_files_to_delete:\n for photo_id, file_id in m2m_photo_files_to_delete:\n PhotoFiles.objects.filter(photo_id=photo_id, file_id=file_id).delete()\n \n if photos_to_update_main:\n Photo.objects.bulk_update(photos_to_update_main, ['main_file'])\n \n if m2m_embedded_to_create:\n FileEmbeddedMedia.objects.bulk_create(m2m_embedded_to_create, ignore_conflicts=True)\n \n if m2m_embedded_to_delete:\n for from_id, to_id in m2m_embedded_to_delete:\n FileEmbeddedMedia.objects.filter(from_file_id=from_id, to_file_id=to_id).delete()\n \n # Delete duplicate files in bulk\n delete_hashes = [f.hash for f in delete_files]\n File.objects.filter(hash__in=delete_hashes).delete()\n deleted_count += len(delete_files)\n \n processed += 1\n if processed % 100 == 0:\n print(f\" Processed {processed}/{total_count} duplicate paths ({100*processed//total_count}%)\")\n \n print(f\"Completed deduplication. Deleted {deleted_count} duplicate files.\")\n\n\ndef reverse_deduplicate(apps, schema_editor):\n \"\"\"\n Reverse migration is a no-op since we can't restore deleted duplicates.\n The unique constraint will be dropped by the AlterField reverse.\n \"\"\"\n pass\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0113_alter_photostack_stack_type'),\n ]\n\n operations = [\n # First, deduplicate existing paths\n migrations.RunPython(\n deduplicate_file_paths,\n reverse_deduplicate,\n ),\n # Then add the unique constraint\n migrations.AlterField(\n model_name='file',\n name='path',\n field=models.TextField(blank=True, default=\"\", unique=True),\n ),\n ]\n" }, { "path": "api/migrations/0115_cleanup_duplicate_photos.py", "content": "# Generated migration to cleanup duplicate Photo records\n# This migration handles Photos with the same image_hash for the same owner\n\nfrom django.db import migrations, transaction\nfrom django.db.models import Count, Case, When, Value, F, IntegerField\n\n\nBATCH_SIZE = 100\n\n\ndef cleanup_duplicate_photos(apps, schema_editor):\n \"\"\"\n Cleanup Photo records that have the same image_hash for the same owner.\n \n Strategy:\n 1. Find all (image_hash, owner) combinations that have multiple Photo records\n 2. For each duplicate group, keep the \"best\" Photo:\n - Prefer non-removed photos\n - Prefer non-trashed photos\n - Prefer photos with main_file\n - Prefer photos with more files attached\n - Prefer photos with more metadata (faces, albums, etc.)\n - Prefer older photos (smaller added_on)\n 3. Merge associations from duplicate Photos to kept Photo:\n - files (M2M)\n - albums (album_user, album_thing, album_place, album_date)\n - faces\n - stacks\n - duplicates (duplicate groups)\n - shared_to\n 4. Delete duplicate Photos\n \n Optimized for large datasets:\n - Uses database annotations to compute scores instead of per-photo queries\n - Uses bulk M2M operations via through model\n - Uses prefetch_related to eliminate N+1 queries\n - Processes in batches with progress logging\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n Face = apps.get_model('api', 'Face')\n AlbumUser = apps.get_model('api', 'AlbumUser')\n \n # Get through models for bulk M2M operations\n PhotoFiles = Photo.files.through\n PhotoStacks = Photo.stacks.through\n PhotoDuplicates = Photo.duplicates.through\n PhotoSharedTo = Photo.shared_to.through\n AlbumUserPhotos = AlbumUser.photos.through\n \n # Find (image_hash, owner) combinations with duplicates\n # Exclude already removed photos from consideration\n duplicate_groups = list(\n Photo.objects\n .filter(removed=False)\n .values('image_hash', 'owner')\n .annotate(count=Count('id'))\n .filter(count__gt=1)\n )\n \n total_count = len(duplicate_groups)\n if total_count == 0:\n print(\"No duplicate photos to clean up.\")\n return\n \n print(f\"Cleaning up {total_count} duplicate photo groups...\")\n \n merged_count = 0\n processed = 0\n \n for dup in duplicate_groups:\n image_hash = dup['image_hash']\n owner_id = dup['owner']\n \n # Get all non-removed Photos with this hash/owner\n # Annotate with scores computed in database\n photos = list(\n Photo.objects\n .filter(image_hash=image_hash, owner_id=owner_id, removed=False)\n .annotate(\n file_count=Count('files', distinct=True),\n face_count=Count('face', distinct=True),\n album_count=Count('albumuser', distinct=True),\n stack_count=Count('stacks', distinct=True),\n # Compute score in database\n score=(\n Case(When(removed=False, then=Value(1000)), default=Value(0), output_field=IntegerField()) +\n Case(When(in_trashcan=False, then=Value(500)), default=Value(0), output_field=IntegerField()) +\n Case(When(main_file__isnull=False, then=Value(200)), default=Value(0), output_field=IntegerField()) +\n F('file_count') * 10 +\n F('face_count') * 5 +\n F('album_count') * 2 +\n F('stack_count') * 2 +\n Case(When(clip_embeddings__isnull=False, then=Value(50)), default=Value(0), output_field=IntegerField()) +\n Case(When(perceptual_hash__isnull=False, then=Value(30)), default=Value(0), output_field=IntegerField()) +\n Case(When(geolocation_json__isnull=False, then=Value(20)), default=Value(0), output_field=IntegerField())\n )\n )\n .select_related('main_file')\n .prefetch_related('files', 'stacks', 'duplicates', 'shared_to', 'albumuser_set')\n .order_by('-score', 'added_on') # Higher score first, then older\n )\n \n if len(photos) <= 1:\n continue\n \n keep_photo = photos[0]\n merge_photos = photos[1:]\n merge_ids = [p.pk for p in merge_photos]\n \n with transaction.atomic():\n # Bulk update faces - single query\n Face.objects.filter(photo_id__in=merge_ids).update(photo=keep_photo)\n \n # Collect existing M2M IDs to prevent duplicates\n existing_file_ids = set(keep_photo.files.values_list('hash', flat=True))\n existing_stack_ids = set(keep_photo.stacks.values_list('id', flat=True))\n existing_duplicate_ids = set(keep_photo.duplicates.values_list('id', flat=True))\n existing_shared_ids = set(keep_photo.shared_to.values_list('id', flat=True))\n existing_album_ids = set(keep_photo.albumuser_set.values_list('id', flat=True))\n \n # Collect M2M entries to create\n new_files = []\n new_stacks = []\n new_duplicates = []\n new_shared = []\n new_albums = []\n \n # Track if we need to update main_file\n main_file_candidate = None\n \n for merge_photo in merge_photos:\n # Collect files (prefetched)\n for file in merge_photo.files.all():\n if file.hash not in existing_file_ids:\n new_files.append(PhotoFiles(photo_id=keep_photo.pk, file_id=file.hash))\n existing_file_ids.add(file.hash)\n \n # If kept photo has no main_file but merge_photo does, remember it\n if not keep_photo.main_file_id and merge_photo.main_file_id and not main_file_candidate:\n main_file_candidate = merge_photo.main_file\n \n # Collect stacks (prefetched)\n for stack in merge_photo.stacks.all():\n if stack.id not in existing_stack_ids:\n new_stacks.append(PhotoStacks(photo_id=keep_photo.pk, photostack_id=stack.id))\n existing_stack_ids.add(stack.id)\n \n # Collect duplicates (prefetched)\n for dup_group in merge_photo.duplicates.all():\n if dup_group.id not in existing_duplicate_ids:\n new_duplicates.append(PhotoDuplicates(photo_id=keep_photo.pk, duplicate_id=dup_group.id))\n existing_duplicate_ids.add(dup_group.id)\n \n # Collect shared_to (prefetched)\n for user in merge_photo.shared_to.all():\n if user.id not in existing_shared_ids:\n new_shared.append(PhotoSharedTo(photo_id=keep_photo.pk, user_id=user.id))\n existing_shared_ids.add(user.id)\n \n # Collect album memberships (prefetched)\n for album in merge_photo.albumuser_set.all():\n if album.id not in existing_album_ids:\n new_albums.append(AlbumUserPhotos(albumuser_id=album.id, photo_id=keep_photo.pk))\n existing_album_ids.add(album.id)\n \n # Bulk create all M2M relationships\n if new_files:\n PhotoFiles.objects.bulk_create(new_files, ignore_conflicts=True)\n if new_stacks:\n PhotoStacks.objects.bulk_create(new_stacks, ignore_conflicts=True)\n if new_duplicates:\n PhotoDuplicates.objects.bulk_create(new_duplicates, ignore_conflicts=True)\n if new_shared:\n PhotoSharedTo.objects.bulk_create(new_shared, ignore_conflicts=True)\n if new_albums:\n AlbumUserPhotos.objects.bulk_create(new_albums, ignore_conflicts=True)\n \n # Update main_file if needed\n if main_file_candidate:\n keep_photo.main_file = main_file_candidate\n keep_photo.save(update_fields=['main_file'])\n \n # Bulk delete merge photos\n # Django's delete() on queryset handles M2M clearing automatically\n Photo.objects.filter(pk__in=merge_ids).delete()\n merged_count += len(merge_ids)\n \n processed += 1\n if processed % BATCH_SIZE == 0:\n print(f\" Processed {processed}/{total_count} duplicate groups ({100*processed//total_count}%)\")\n \n if merged_count > 0:\n print(f\"Completed cleanup. Deleted {merged_count} duplicate Photo records.\")\n\n\ndef reverse_cleanup(apps, schema_editor):\n \"\"\"\n Reverse migration is a no-op since deleted photos cannot be restored.\n \"\"\"\n pass\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0114_add_file_path_unique'),\n ]\n\n operations = [\n migrations.RunPython(\n cleanup_duplicate_photos,\n reverse_cleanup,\n ),\n ]\n" }, { "path": "api/migrations/0116_cleanup_duplicate_groups_removed_photos.py", "content": "# Generated migration to clean up Duplicate groups containing removed photos\n# This removes removed=True photos from groups and deletes groups with 0-1 photos remaining\n\nfrom django.db import migrations\n\n\ndef cleanup_duplicate_groups(apps, schema_editor):\n \"\"\"\n Remove removed=True photos from Duplicate groups.\n Delete Duplicate groups that end up with 0 or 1 photos.\n \n This is needed because migration 0115 marks duplicate Photos as removed=True\n but doesn't remove them from their Duplicate group M2M relationships.\n \"\"\"\n Duplicate = apps.get_model('api', 'Duplicate')\n \n cleaned_count = 0\n deleted_count = 0\n \n for duplicate in Duplicate.objects.all():\n # Get removed photos in this group\n removed_photos = duplicate.photos.filter(removed=True)\n removed_count = removed_photos.count()\n \n if removed_count > 0:\n # Remove the removed photos from the group\n for photo in removed_photos:\n duplicate.photos.remove(photo)\n cleaned_count += removed_count\n \n # Check if group now has 0 or 1 photos (no longer a valid duplicate group)\n remaining = duplicate.photos.filter(removed=False).count()\n if remaining <= 1:\n # Clear remaining photos first to avoid orphan M2M entries\n duplicate.photos.clear()\n duplicate.delete()\n deleted_count += 1\n \n if cleaned_count or deleted_count:\n print(f\"Cleaned {cleaned_count} removed photos from duplicate groups\")\n print(f\"Deleted {deleted_count} empty/single-photo duplicate groups\")\n\n\ndef reverse_cleanup(apps, schema_editor):\n \"\"\"\n Reverse migration is a no-op since we can't restore removed photos to groups.\n The photos still exist (just marked removed=True) but we don't track which\n groups they belonged to.\n \"\"\"\n pass\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0115_cleanup_duplicate_photos'),\n ]\n\n operations = [\n migrations.RunPython(cleanup_duplicate_groups, reverse_cleanup),\n ]\n" }, { "path": "api/migrations/0117_delete_removed_photos.py", "content": "# Generated migration to delete removed photos\n# These are duplicate photos that were merged in migration 0115\n\nfrom django.db import migrations\n\n\ndef delete_removed_photos(apps, schema_editor):\n \"\"\"\n Delete all Photo records marked as removed=True.\n \n These are duplicate photos that were already merged in migration 0115.\n Their relationships (faces, albums, files, stacks, duplicates) were\n reassigned to the kept photo, so these are now orphan records.\n \n Deleting them is cleaner than soft-delete because:\n 1. No need to filter removed=True everywhere in queries\n 2. No orphan data cluttering the database\n 3. Clearer data model\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n \n # Find all removed photos\n removed_photos = Photo.objects.filter(removed=True)\n count = removed_photos.count()\n \n if count > 0:\n # Delete them - relationships were already cleared/reassigned in 0115\n removed_photos.delete()\n print(f\"Deleted {count} removed (duplicate) photos\")\n\n\ndef reverse_delete(apps, schema_editor):\n \"\"\"\n Reverse migration is a no-op - deleted photos cannot be restored.\n \"\"\"\n pass\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0116_cleanup_duplicate_groups_removed_photos'),\n ]\n\n operations = [\n migrations.RunPython(delete_removed_photos, reverse_delete),\n ]\n" }, { "path": "api/migrations/0118_alter_longrunningjob_job_type.py", "content": "# Generated by Django 5.2.9 on 2026-01-28 13:43\n\nfrom django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0117_delete_removed_photos'),\n ]\n\n operations = [\n migrations.AlterField(\n model_name='longrunningjob',\n name='job_type',\n field=models.PositiveIntegerField(choices=[(1, 'Scan Photos'), (2, 'Generate Event Albums'), (3, 'Regenerate Event Titles'), (4, 'Train Faces'), (5, 'Delete Missing Photos'), (7, 'Scan Faces'), (6, 'Calculate Clip Embeddings'), (8, 'Find Similar Faces'), (9, 'Download Selected Photos'), (10, 'Download Models'), (11, 'Add Geolocation'), (12, 'Generate Tags'), (13, 'Generate Face Embeddings'), (14, 'Scan Missing Photos'), (15, 'Detect Duplicate Photos'), (16, 'Repair File Variants')]),\n ),\n ]\n" }, { "path": "api/migrations/0119_add_public_sharing_options.py", "content": "# Generated migration for public sharing options\n\nfrom django.db import migrations, models\nimport api.models.user\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0118_alter_longrunningjob_job_type\"),\n ]\n\n operations = [\n # Add public_sharing_defaults to User model\n migrations.AddField(\n model_name=\"user\",\n name=\"public_sharing_defaults\",\n field=models.JSONField(default=api.models.user.get_default_public_sharing_settings),\n ),\n # Add sharing option fields to AlbumUserShare model\n migrations.AddField(\n model_name=\"albumusershare\",\n name=\"share_location\",\n field=models.BooleanField(blank=True, default=None, null=True),\n ),\n migrations.AddField(\n model_name=\"albumusershare\",\n name=\"share_camera_info\",\n field=models.BooleanField(blank=True, default=None, null=True),\n ),\n migrations.AddField(\n model_name=\"albumusershare\",\n name=\"share_timestamps\",\n field=models.BooleanField(blank=True, default=None, null=True),\n ),\n migrations.AddField(\n model_name=\"albumusershare\",\n name=\"share_captions\",\n field=models.BooleanField(blank=True, default=None, null=True),\n ),\n migrations.AddField(\n model_name=\"albumusershare\",\n name=\"share_faces\",\n field=models.BooleanField(blank=True, default=None, null=True),\n ),\n ]\n" }, { "path": "api/migrations/0120_rename_thumbnails_uuid_to_hash.py", "content": "# Migration to rename thumbnail files from UUID to image_hash\n# This migration fixes the thumbnail naming issue where thumbnails were being\n# created with UUID names but the frontend expected image_hash names\n\nimport os\nfrom django.conf import settings\nfrom django.db import migrations\n\n\ndef rename_thumbnails_uuid_to_hash(apps, schema_editor):\n \"\"\"\n Rename existing thumbnail files from UUID-based names to image_hash-based names.\n This only renames files that exist and updates the database records.\n Uses batch processing for improved performance with large photo collections.\n \"\"\"\n Photo = apps.get_model('api', 'Photo')\n Thumbnail = apps.get_model('api', 'Thumbnail')\n \n BATCH_SIZE = 1000 # Process thumbnails in batches of 1000\n \n # Get total count for progress reporting\n total_count = Thumbnail.objects.count()\n print(f\"Starting thumbnail migration for {total_count} photos...\")\n \n renamed_count = 0\n skipped_count = 0\n processed_count = 0\n \n # Process thumbnails in batches to avoid loading all into memory at once\n thumbnails_to_update = []\n \n # Use iterator() to avoid loading all objects into memory\n # Process in batches using only() to load only required fields\n for thumbnail in Thumbnail.objects.select_related('photo').only(\n 'photo_id', 'photo__id', 'photo__image_hash', 'photo__video',\n 'thumbnail_big', 'square_thumbnail', 'square_thumbnail_small'\n ).iterator(chunk_size=BATCH_SIZE):\n photo = thumbnail.photo\n photo_uuid = str(photo.id)\n photo_hash = photo.image_hash\n \n # Skip if UUID and hash are the same (shouldn't happen, but be safe)\n if photo_uuid == photo_hash:\n skipped_count += 1\n processed_count += 1\n continue\n \n # Process each thumbnail type\n thumbnail_types = [\n ('thumbnails_big', '.webp', False), # (path, extension, is_video)\n ('square_thumbnails', '.webp' if not photo.video else '.mp4', photo.video),\n ('square_thumbnails_small', '.webp' if not photo.video else '.mp4', photo.video),\n ]\n \n needs_update = False\n \n for thumb_dir, ext, _ in thumbnail_types:\n old_path = os.path.join(settings.MEDIA_ROOT, thumb_dir, f\"{photo_uuid}{ext}\")\n new_path = os.path.join(settings.MEDIA_ROOT, thumb_dir, f\"{photo_hash}{ext}\")\n \n # Only rename if old file exists and new file doesn't\n if os.path.exists(old_path) and not os.path.exists(new_path):\n try:\n os.rename(old_path, new_path)\n needs_update = True\n except Exception as e:\n print(f\"Warning: Could not rename {old_path} to {new_path}: {e}\")\n \n # Queue for batch update if any files were renamed\n if needs_update:\n filetype = '.mp4' if photo.video else '.webp'\n thumbnail.thumbnail_big = os.path.join('thumbnails_big', f\"{photo_hash}.webp\")\n thumbnail.square_thumbnail = os.path.join('square_thumbnails', f\"{photo_hash}{filetype}\")\n thumbnail.square_thumbnail_small = os.path.join('square_thumbnails_small', f\"{photo_hash}{filetype}\")\n thumbnails_to_update.append(thumbnail)\n renamed_count += 1\n else:\n skipped_count += 1\n \n processed_count += 1\n \n # Batch update every BATCH_SIZE records\n if len(thumbnails_to_update) >= BATCH_SIZE:\n Thumbnail.objects.bulk_update(\n thumbnails_to_update,\n ['thumbnail_big', 'square_thumbnail', 'square_thumbnail_small'],\n batch_size=BATCH_SIZE\n )\n print(f\"Progress: {processed_count}/{total_count} processed, {renamed_count} renamed, {skipped_count} skipped\")\n thumbnails_to_update = []\n \n # Update any remaining thumbnails in the final batch\n if thumbnails_to_update:\n Thumbnail.objects.bulk_update(\n thumbnails_to_update,\n ['thumbnail_big', 'square_thumbnail', 'square_thumbnail_small'],\n batch_size=BATCH_SIZE\n )\n \n print(f\"Migration complete: {renamed_count} photos renamed, {skipped_count} photos skipped\")\n\n\ndef reverse_rename_thumbnails(apps, schema_editor):\n \"\"\"\n This migration cannot be easily reversed because we would need to know\n the original UUID for each photo. The forward migration renames files\n from UUID to image_hash, but reversing would require knowing which UUID\n was used originally, which we don't store.\n \"\"\"\n print(\"Warning: This migration cannot be reversed. Thumbnails will keep image_hash names.\")\n print(\"If you need to revert, regenerate thumbnails from scratch.\")\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n ('api', '0119_add_public_sharing_options'),\n ]\n\n operations = [\n migrations.RunPython(\n rename_thumbnails_uuid_to_hash,\n reverse_rename_thumbnails,\n ),\n ]\n" }, { "path": "api/migrations/0121_add_default_tagging_model.py", "content": "\"\"\"\nMigration to add a default TAGGING_MODEL entry to the constance database.\n\nWhen TAGGING_MODEL was added to CONSTANCE_CONFIG, existing systems that upgraded\nwould not have this key in the constance database backend. While constance normally\nfalls back to the default from CONSTANCE_CONFIG, this migration explicitly sets the\ndefault to ensure compatibility with old systems.\n\"\"\"\n\nfrom django.db import migrations\n\n\ndef add_default_tagging_model(apps, schema_editor):\n \"\"\"Add TAGGING_MODEL default value to constance DB if it doesn't exist.\"\"\"\n try:\n Constance = apps.get_model(\"constance\", \"Constance\")\n if not Constance.objects.filter(key=\"TAGGING_MODEL\").exists():\n Constance.objects.create(key=\"TAGGING_MODEL\", value='\"places365\"')\n except LookupError:\n # constance model not available, skip\n pass\n\n\ndef reverse_migration(apps, schema_editor):\n \"\"\"Remove TAGGING_MODEL from constance DB if it has the default value.\"\"\"\n try:\n Constance = apps.get_model(\"constance\", \"Constance\")\n Constance.objects.filter(key=\"TAGGING_MODEL\", value='\"places365\"').delete()\n except LookupError:\n pass\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0120_rename_thumbnails_uuid_to_hash\"),\n ]\n\n operations = [\n migrations.RunPython(add_default_tagging_model, reverse_migration),\n ]\n" }, { "path": "api/migrations/0121_user_save_face_tags_to_disk.py", "content": "from django.db import migrations, models\n\n\nclass Migration(migrations.Migration):\n\n dependencies = [\n (\"api\", \"0120_rename_thumbnails_uuid_to_hash\"),\n ]\n\n operations = [\n migrations.AddField(\n model_name=\"user\",\n name=\"save_face_tags_to_disk\",\n field=models.BooleanField(default=False),\n ),\n ]\n" }, { "path": "api/migrations/__init__.py", "content": "" }, { "path": "api/ml_models.py", "content": "import math\nimport os\nimport tarfile\nfrom pathlib import Path\n\nimport requests\nfrom constance import config as site_config\nfrom django.conf import settings\n\nfrom api import util\nfrom api.models.long_running_job import LongRunningJob\n\n\nclass MlTypes:\n CAPTIONING = \"captioning\"\n FACE_RECOGNITION = \"face_recognition\"\n CATEGORIES = \"categories\"\n CLIP = \"clip\"\n LLM = \"llm\"\n MOONDREAM = \"moondream\"\n TAGGING = \"tagging\"\n\n\nML_MODELS = [\n {\n \"id\": 1,\n \"name\": \"im2txt\",\n \"url\": \"https://github.com/LibrePhotos/librephotos-docker/releases/download/0.1/im2txt.tar.gz\",\n \"type\": MlTypes.CAPTIONING,\n \"unpack-command\": \"tar -zxC\",\n \"target-dir\": \"im2txt\",\n },\n {\n \"id\": 2,\n \"name\": \"clip-embeddings\",\n \"url\": \"https://github.com/LibrePhotos/librephotos-docker/releases/download/0.1/clip-embeddings.tar.gz\",\n \"type\": MlTypes.CLIP,\n \"unpack-command\": \"tar -zxC\",\n \"target-dir\": \"clip-embeddings\",\n },\n {\n \"id\": 3,\n \"name\": \"places365\",\n \"url\": \"https://github.com/LibrePhotos/librephotos-docker/releases/download/0.1/places365.tar.gz\",\n \"type\": MlTypes.CATEGORIES,\n \"unpack-command\": \"tar -zxC\",\n \"target-dir\": \"places365\",\n },\n {\n \"id\": 4,\n \"name\": \"resnet18\",\n \"url\": \"https://download.pytorch.org/models/resnet18-5c106cde.pth\",\n \"type\": MlTypes.CATEGORIES,\n \"unpack-command\": None,\n \"target-dir\": \"resnet18-5c106cde.pth\",\n },\n {\n \"id\": 6,\n \"name\": \"blip_base_capfilt_large\",\n \"url\": \"https://huggingface.co/derneuere/librephotos_models/resolve/main/blip_large.tar.gz?download=true\",\n \"type\": MlTypes.CAPTIONING,\n \"unpack-command\": \"tar -zxC\",\n \"target-dir\": \"blip\",\n },\n {\n \"id\": 8,\n \"name\": \"mistral-7b-instruct-v0.2.Q5_K_M\",\n \"url\": \"https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-GGUF/resolve/main/mistral-7b-instruct-v0.2.Q5_K_M.gguf?download=true\",\n \"type\": MlTypes.LLM,\n \"unpack-command\": None,\n \"target-dir\": \"mistral-7b-instruct-v0.2.Q5_K_M.gguf\",\n },\n {\n \"id\": 11,\n \"name\": \"siglip2\",\n \"url\": \"https://huggingface.co/onnx-community/siglip2-base-patch16-384-ONNX/resolve/main/onnx/vision_model.onnx\",\n \"type\": MlTypes.TAGGING,\n \"unpack-command\": None,\n \"target-dir\": \"siglip2/vision_model.onnx\",\n \"additional_files\": [\n {\n \"url\": \"https://huggingface.co/onnx-community/siglip2-base-patch16-384-ONNX/resolve/main/onnx/text_model.onnx\",\n \"target\": \"siglip2/text_model.onnx\",\n },\n {\n \"url\": \"https://huggingface.co/onnx-community/siglip2-base-patch16-384-ONNX/resolve/main/tokenizer.model\",\n \"target\": \"siglip2/tokenizer.model\",\n },\n ],\n },\n {\n # Moondream 2 GGUF model for llama-cpp-python multimodal support\n \"id\": 9,\n \"name\": \"moondream\",\n \"url\": \"https://huggingface.co/moondream/moondream-2b-2025-04-14-4bit/resolve/main/moondream2-text-model-f16.gguf?download=true\",\n \"type\": MlTypes.MOONDREAM,\n \"unpack-command\": None,\n \"target-dir\": \"moondream2-text-model-f16.gguf\",\n \"additional_files\": [\n {\n \"url\": \"https://huggingface.co/moondream/moondream-2b-2025-04-14-4bit/resolve/main/moondream2-mmproj-f16.gguf?download=true\",\n \"target\": \"moondream2-mmproj-f16.gguf\",\n }\n ],\n },\n]\n\n\ndef download_model(model):\n model = model.copy()\n if model[\"type\"] == MlTypes.LLM:\n util.logger.info(\"Downloading LLM model\")\n model_to_download = site_config.LLM_MODEL\n if not model_to_download or str(model_to_download).strip().lower() == \"none\":\n util.logger.info(\"No LLM model selected\")\n return\n util.logger.info(f\"Model to download: {model_to_download}\")\n # Look through ML_MODELS and find the model with the name\n for ml_model in ML_MODELS:\n if ml_model[\"name\"] == model_to_download:\n model = ml_model\n elif model[\"type\"] == MlTypes.MOONDREAM:\n util.logger.info(\"Downloading Moondream model\")\n model_to_download = site_config.LLM_MODEL\n if model_to_download != \"moondream\":\n util.logger.info(\"Moondream not selected\")\n return\n util.logger.info(f\"Model to download: {model_to_download}\")\n # Look through ML_MODELS and find the model with the name\n for ml_model in ML_MODELS:\n if ml_model[\"name\"] == model_to_download:\n model = ml_model\n elif model[\"type\"] == MlTypes.CAPTIONING:\n util.logger.info(\"Downloading captioning model\")\n model_to_download = site_config.CAPTIONING_MODEL\n util.logger.info(f\"Model to download: {model_to_download}\")\n # Look through ML_MODELS and find the model with the name\n for ml_model in ML_MODELS:\n if ml_model[\"name\"] == model_to_download:\n model = ml_model\n elif model[\"type\"] == MlTypes.TAGGING:\n util.logger.info(\"Downloading tagging model\")\n model_to_download = site_config.TAGGING_MODEL\n if model_to_download != model[\"name\"]:\n util.logger.info(\n f\"Tagging model {model['name']} not selected (current: {model_to_download})\"\n )\n return\n util.logger.info(f\"Model to download: {model_to_download}\")\n\n util.logger.info(f\"Downloading model {model['name']}\")\n model_folder = Path(settings.MEDIA_ROOT) / \"data_models\"\n\n # Handle regular models\n target_dir = model_folder / model[\"target-dir\"]\n\n if target_dir.exists():\n util.logger.info(f\"Model {model['name']} already downloaded\")\n # Check if all additional files exist for models like Moondream\n if model.get(\"additional_files\"):\n for additional_file in model[\"additional_files\"]:\n additional_target = model_folder / additional_file[\"target\"]\n if not additional_target.exists():\n util.logger.info(\n f\"Additional file {additional_file['target']} missing, downloading...\"\n )\n _download_file(\n additional_file[\"url\"],\n additional_target,\n f\"{model['name']} ({additional_file['target']})\",\n )\n return\n\n if model[\"unpack-command\"] == \"tar -zxC\":\n target_dir = model_folder / (model[\"target-dir\"] + \".tar.gz\")\n if model[\"unpack-command\"] == \"tar -xvf\":\n target_dir = model_folder / (model[\"target-dir\"] + \".tar\")\n if model[\"unpack-command\"] is None:\n target_dir = model_folder / model[\"target-dir\"]\n\n _download_file(model[\"url\"], target_dir, model[\"name\"])\n\n if model[\"unpack-command\"] == \"tar -zxC\":\n with tarfile.open(target_dir, mode=\"r:gz\") as tar:\n tar.extractall(path=model_folder)\n os.remove(target_dir)\n if model[\"unpack-command\"] == \"tar -xvf\":\n with tarfile.open(target_dir, mode=\"r:\") as tar:\n tar.extractall(path=model_folder)\n os.remove(target_dir)\n\n # Download additional files if they exist (e.g., mmproj for Moondream)\n if model.get(\"additional_files\"):\n for additional_file in model[\"additional_files\"]:\n additional_target = model_folder / additional_file[\"target\"]\n if not additional_target.exists():\n _download_file(\n additional_file[\"url\"],\n additional_target,\n f\"{model['name']} ({additional_file['target']})\",\n )\n\n\ndef _download_file(url, target_path, model_name):\n \"\"\"Helper function to download a single file with progress tracking\"\"\"\n target_path = Path(target_path)\n target_path.parent.mkdir(parents=True, exist_ok=True)\n response = requests.get(url, stream=True, allow_redirects=True)\n total_size = int(response.headers.get(\"content-length\", 0))\n block_size = 1024\n current_progress = 0\n previous_percentage = -1\n\n with open(target_path, \"wb\") as target_file:\n for chunk in response.iter_content(chunk_size=block_size):\n if chunk:\n target_file.write(chunk)\n current_progress += len(chunk)\n\n if total_size > 0:\n percentage = math.floor((current_progress / total_size) * 100)\n\n if percentage != previous_percentage:\n util.logger.info(\n f\"Downloading {model_name}: {current_progress}/{total_size} ({percentage}%)\"\n )\n previous_percentage = percentage\n\n if total_size == 0:\n util.logger.info(\n f\"Downloaded {model_name}: {current_progress} bytes (size unknown during transfer)\"\n )\n\n\ndef download_models(user):\n lrj = LongRunningJob.create_job(\n user=user,\n job_type=LongRunningJob.JOB_DOWNLOAD_MODELS,\n start_now=True,\n )\n lrj.update_progress(current=0, target=len(ML_MODELS))\n\n model_folder = Path(settings.MEDIA_ROOT) / \"data_models\"\n model_folder.mkdir(parents=True, exist_ok=True)\n\n for idx, model in enumerate(ML_MODELS):\n download_model(model)\n lrj.update_progress(current=idx + 1)\n\n lrj.complete()\n\n\ndef do_all_models_exist():\n model_folder = Path(settings.MEDIA_ROOT) / \"data_models\"\n for model in ML_MODELS:\n if model[\"type\"] in (MlTypes.LLM, MlTypes.MOONDREAM, MlTypes.TAGGING):\n if not model and model != \"none\":\n continue\n\n # Check main model file\n target_dir = model_folder / model[\"target-dir\"]\n if not target_dir.exists():\n return False\n\n # Check additional files if they exist (like mmproj for Moondream)\n if model.get(\"additional_files\"):\n for additional_file in model[\"additional_files\"]:\n additional_target = model_folder / additional_file[\"target\"]\n if not additional_target.exists():\n return False\n return True\n" }, { "path": "api/models/__init__.py", "content": "from api.models.album_auto import AlbumAuto\nfrom api.models.album_date import AlbumDate\nfrom api.models.album_place import AlbumPlace\nfrom api.models.album_thing import AlbumThing\nfrom api.models.album_user import AlbumUser\nfrom api.models.cluster import Cluster\nfrom api.models.duplicate import Duplicate\nfrom api.models.face import Face\nfrom api.models.file import File\nfrom api.models.long_running_job import LongRunningJob\nfrom api.models.person import Person\nfrom api.models.photo import Photo\nfrom api.models.photo_caption import PhotoCaption\nfrom api.models.photo_metadata import MetadataEdit, MetadataFile, PhotoMetadata\nfrom api.models.photo_search import PhotoSearch\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.stack_review import StackReview\nfrom api.models.thumbnail import Thumbnail\nfrom api.models.user import User\n\n__all__ = [\n \"AlbumAuto\",\n \"AlbumDate\",\n \"AlbumPlace\",\n \"AlbumThing\",\n \"AlbumUser\",\n \"Cluster\",\n \"Duplicate\",\n \"Face\",\n \"LongRunningJob\",\n \"MetadataEdit\",\n \"MetadataFile\",\n \"Person\",\n \"Photo\",\n \"PhotoCaption\",\n \"PhotoMetadata\",\n \"PhotoSearch\",\n \"PhotoStack\",\n \"StackReview\",\n \"Thumbnail\",\n \"User\",\n \"File\",\n]\n" }, { "path": "api/models/album_auto.py", "content": "from collections import Counter\n\nfrom django.db import models\n\nfrom api import util\nfrom api.models.person import Person\nfrom api.models.photo import Photo\nfrom api.models.user import User, get_deleted_user\n\n\nclass AlbumAuto(models.Model):\n title = models.CharField(\n blank=False, null=False, max_length=512, default=\"Untitled Album\"\n )\n timestamp = models.DateTimeField(db_index=True)\n created_on = models.DateTimeField(auto_now=False, db_index=True)\n gps_lat = models.FloatField(blank=True, null=True)\n gps_lon = models.FloatField(blank=True, null=True)\n photos = models.ManyToManyField(Photo)\n favorited = models.BooleanField(default=False, db_index=True)\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n\n shared_to = models.ManyToManyField(User, related_name=\"album_auto_shared_to\")\n\n class Meta:\n unique_together = (\"timestamp\", \"owner\")\n\n def _generate_title(self):\n try:\n weekday = \"\"\n time = \"\"\n loc = \"\"\n if self.timestamp:\n weekday = util.weekdays[self.timestamp.isoweekday()]\n hour = self.timestamp.hour\n if hour > 0 and hour < 5:\n time = \"Early Morning\"\n elif hour >= 5 and hour < 12:\n time = \"Morning\"\n elif hour >= 12 and hour < 18:\n time = \"Afternoon\"\n elif hour >= 18 and hour <= 24:\n time = \"Evening\"\n\n when = \" \".join([weekday, time])\n\n photos = self.photos.all()\n\n loc = \"\"\n pep = \"\"\n\n places = []\n people = []\n timestamps = []\n\n for photo in photos:\n if (\n photo.geolocation_json\n and \"places\" in photo.geolocation_json.keys()\n and len(photo.geolocation_json[\"places\"]) > 0\n ):\n places = photo.geolocation_json[\"places\"]\n\n timestamps.append(photo.exif_timestamp)\n\n faces = photo.faces.all()\n for face in faces:\n people.append(face.person.name)\n\n if len(places) > 0:\n cnts_places = Counter(places)\n loc = \"in \" + \" and \".join(dict(cnts_places.most_common(2)).keys())\n if len(people) > 0:\n cnts_people = Counter(people)\n names = dict(\n [\n (k, v)\n for k, v in cnts_people.most_common(2)\n if k.lower() != \"unknown\"\n and k.lower() != Person.UNKNOWN_PERSON_NAME\n ]\n ).keys()\n if len(names) > 0:\n pep = \"with \" + \" and \".join(names)\n if len(timestamps) > 0:\n if (max(timestamps) - min(timestamps)).days >= 3:\n when = \"%d days\" % ((max(timestamps) - min(timestamps)).days)\n\n weekend = [5, 6]\n if (\n max(timestamps).weekday() in weekend\n and min(timestamps).weekday() in weekend\n and not (max(timestamps).weekday() == min(timestamps).weekday())\n ):\n when = \"Weekend\"\n\n title = \" \".join([when, pep, loc]).strip()\n # Ensure title is never empty\n if not title:\n title = f\"Album from {self.timestamp.strftime('%Y-%m-%d')}\"\n self.title = title\n self.save()\n except Exception as e:\n util.logger.exception(e)\n # Set a fallback title if something goes wrong\n self.title = f\"Album from {self.timestamp.strftime('%Y-%m-%d')}\"\n self.save()\n\n def __str__(self):\n return \"%d: %s\" % (self.id, self.title)\n" }, { "path": "api/models/album_date.py", "content": "from django.db import models\n\nfrom api.models.photo import Photo\nfrom api.models.user import User, get_deleted_user\n\n\nclass AlbumDate(models.Model):\n title = models.CharField(blank=True, default=\"\", max_length=512, db_index=True)\n date = models.DateField(db_index=True, null=True)\n photos = models.ManyToManyField(Photo)\n favorited = models.BooleanField(default=False, db_index=True)\n location = models.JSONField(blank=True, db_index=True, null=True)\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n shared_to = models.ManyToManyField(User, related_name=\"album_date_shared_to\")\n objects = models.Manager()\n\n class Meta:\n unique_together = (\"date\", \"owner\")\n\n def __str__(self):\n return str(self.date) + \" (\" + str(self.owner) + \")\"\n\n def ordered_photos(self):\n return self.photos.all().order_by(\"-exif_timestamp\")\n\n\ndef get_or_create_album_date(date, owner):\n try:\n return AlbumDate.objects.get_or_create(date=date, owner=owner)[0]\n except AlbumDate.MultipleObjectsReturned:\n return AlbumDate.objects.filter(date=date, owner=owner).first()\n\n\ndef get_album_date(date, owner):\n try:\n return AlbumDate.objects.get(date=date, owner=owner)\n except Exception:\n return None\n\n\ndef get_album_nodate(owner):\n return AlbumDate.objects.get_or_create(date=None, owner=owner)[0]\n" }, { "path": "api/models/album_place.py", "content": "from django.db import models\n\nfrom api.models.photo import Photo\nfrom api.models.user import User, get_deleted_user\n\n\nclass AlbumPlace(models.Model):\n title = models.CharField(max_length=512, db_index=True)\n photos = models.ManyToManyField(Photo)\n geolocation_level = models.IntegerField(db_index=True, null=True)\n favorited = models.BooleanField(default=False, db_index=True)\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n\n shared_to = models.ManyToManyField(User, related_name=\"album_place_shared_to\")\n\n class Meta:\n unique_together = (\"title\", \"owner\")\n\n def __str__(self):\n return \"%d: %s\" % (self.id, self.title)\n\n\ndef get_album_place(title, owner):\n return AlbumPlace.objects.get_or_create(title=title, owner=owner)[0]\n" }, { "path": "api/models/album_thing.py", "content": "from django.db import models\nfrom django.db.models.signals import m2m_changed\nfrom django.dispatch import receiver\n\nfrom api.models.photo import Photo\nfrom api.models.user import User, get_deleted_user\n\n\ndef update_default_cover_photo(instance):\n if instance.cover_photos.count() < 4:\n photos_to_add = instance.photos.filter(hidden=False)[\n : 4 - instance.cover_photos.count()\n ]\n instance.cover_photos.add(*photos_to_add)\n\n\nclass AlbumThing(models.Model):\n title = models.CharField(max_length=512, db_index=True)\n photos = models.ManyToManyField(Photo)\n thing_type = models.CharField(max_length=512, db_index=True, null=True)\n favorited = models.BooleanField(default=False, db_index=True)\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n shared_to = models.ManyToManyField(User, related_name=\"album_thing_shared_to\")\n cover_photos = models.ManyToManyField(\n Photo, related_name=\"album_thing_cover_photos\"\n )\n photo_count = models.IntegerField(default=0)\n\n class Meta:\n constraints = [\n models.UniqueConstraint(\n fields=[\"title\", \"thing_type\", \"owner\"], name=\"unique AlbumThing\"\n )\n ]\n\n def save(self, *args, **kwargs):\n super().save(*args, **kwargs)\n\n def update_default_cover_photo(self):\n update_default_cover_photo(self)\n\n def __str__(self):\n return \"%d: %s\" % (self.id or 0, self.title)\n\n\n@receiver(m2m_changed, sender=AlbumThing.photos.through)\ndef update_photo_count(sender, instance, action, reverse, model, pk_set, **kwargs):\n if action == \"post_add\" or (action == \"post_remove\" and not reverse):\n count = instance.photos.filter(hidden=False).count()\n instance.photo_count = count\n instance.save(update_fields=[\"photo_count\"])\n instance.update_default_cover_photo()\n\n\ndef get_album_thing(title, owner, thing_type=None):\n return AlbumThing.objects.get_or_create(\n title=title, owner=owner, thing_type=thing_type\n )[0]\n" }, { "path": "api/models/album_user.py", "content": "from django.db import models\n\nfrom api.models.photo import Photo\nfrom api.models.user import User, get_deleted_user\n\n\nclass AlbumUser(models.Model):\n title = models.CharField(max_length=512)\n created_on = models.DateTimeField(auto_now=True, db_index=True)\n photos = models.ManyToManyField(Photo)\n favorited = models.BooleanField(default=False, db_index=True)\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n cover_photo = models.ForeignKey(\n Photo,\n related_name=\"album_user\",\n on_delete=models.SET_NULL,\n blank=False,\n null=True,\n )\n\n shared_to = models.ManyToManyField(User, related_name=\"album_user_shared_to\")\n\n def __str__(self):\n return f\"{self.title} ({self.owner.username})\"\n\n class Meta:\n unique_together = (\"title\", \"owner\")\n" }, { "path": "api/models/album_user_share.py", "content": "import uuid\nfrom django.db import models\nfrom django.utils import timezone\n\nfrom api.models.album_user import AlbumUser\n\n\nclass AlbumUserShare(models.Model):\n album = models.OneToOneField(\n AlbumUser, on_delete=models.CASCADE, related_name=\"share\"\n )\n enabled = models.BooleanField(default=False, db_index=True)\n slug = models.SlugField(\n max_length=64, unique=True, null=True, blank=True, db_index=True\n )\n expires_at = models.DateTimeField(null=True, blank=True, db_index=True)\n\n # Sharing options - None means inherit from user defaults, True/False overrides\n share_location = models.BooleanField(null=True, blank=True, default=None)\n share_camera_info = models.BooleanField(null=True, blank=True, default=None)\n share_timestamps = models.BooleanField(null=True, blank=True, default=None)\n share_captions = models.BooleanField(null=True, blank=True, default=None)\n share_faces = models.BooleanField(null=True, blank=True, default=None)\n\n def ensure_slug(self) -> None:\n if self.enabled and not self.slug:\n base = uuid.uuid4().hex[:12]\n candidate = base\n idx = 0\n while (\n AlbumUserShare.objects.filter(slug=candidate)\n .exclude(id=self.id)\n .exists()\n ):\n idx += 1\n candidate = f\"{base}-{idx}\"\n self.slug = candidate\n\n def is_active(self) -> bool:\n if not self.enabled:\n return False\n if self.expires_at is None:\n return True\n return self.expires_at >= timezone.now()\n\n def save(self, *args, **kwargs):\n if self.enabled and not self.slug:\n self.ensure_slug()\n super().save(*args, **kwargs)\n\n def get_effective_sharing_settings(self) -> dict:\n \"\"\"Resolve effective sharing settings.\n \n Priority: album override > user defaults > system defaults (all False)\n \"\"\"\n from api.models.user import get_default_public_sharing_settings\n \n # Start with system defaults (all False)\n defaults = get_default_public_sharing_settings()\n \n # Apply user defaults if available\n user_defaults = getattr(self.album.owner, 'public_sharing_defaults', None)\n if user_defaults:\n defaults.update(user_defaults)\n \n # Apply album-level overrides (only non-None values)\n overrides = {\n 'share_location': self.share_location,\n 'share_camera_info': self.share_camera_info,\n 'share_timestamps': self.share_timestamps,\n 'share_captions': self.share_captions,\n 'share_faces': self.share_faces,\n }\n \n for key, value in overrides.items():\n if value is not None:\n defaults[key] = value\n \n return defaults\n" }, { "path": "api/models/cluster.py", "content": "import numpy as np\nfrom django.core.exceptions import MultipleObjectsReturned\nfrom django.db import models\n\nfrom api.models.person import Person\nfrom api.models.user import User, get_deleted_user\nfrom api.util import logger\n\nUNKNOWN_CLUSTER_ID = -1\nUNKNOWN_CLUSTER_NAME = \"Other Unknown Cluster\"\n\n\nclass Cluster(models.Model):\n person = models.ForeignKey(\n Person,\n on_delete=models.SET_NULL,\n related_name=\"clusters\",\n blank=True,\n null=True,\n )\n mean_face_encoding = models.TextField()\n cluster_id = models.IntegerField(null=True)\n name = models.TextField(null=True)\n\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None, null=True\n )\n\n def __str__(self):\n return \"%d\" % self.id\n\n def get_mean_encoding_array(self) -> np.ndarray:\n return np.frombuffer(bytes.fromhex(self.mean_face_encoding))\n\n def set_metadata(self, all_vectors):\n self.mean_face_encoding = (\n Cluster.calculate_mean_face_encoding(all_vectors).tobytes().hex()\n )\n\n @staticmethod\n def get_or_create_cluster_by_name(user: User, name):\n return Cluster.objects.get_or_create(owner=user, name=name)[0]\n\n @staticmethod\n def get_or_create_cluster_by_id(user: User, cluster_id: int):\n try:\n return Cluster.objects.get_or_create(owner=user, cluster_id=cluster_id)[0]\n except MultipleObjectsReturned:\n logger.error(\n \"Multiple clusters found with id %d. Choosing first one\" % cluster_id\n )\n return Cluster.objects.filter(owner=user, cluster_id=cluster_id).first()\n\n @staticmethod\n def calculate_mean_face_encoding(all_encodings):\n return np.mean(a=all_encodings, axis=0, dtype=np.float64)\n\n\ndef get_unknown_cluster(user: User) -> Cluster:\n unknown_cluster: Cluster = Cluster.get_or_create_cluster_by_id(\n user, UNKNOWN_CLUSTER_ID\n )\n if unknown_cluster.person is not None:\n unknown_cluster.person = None\n unknown_cluster.name = UNKNOWN_CLUSTER_NAME\n unknown_cluster.save()\n return unknown_cluster\n" }, { "path": "api/models/duplicate.py", "content": "\"\"\"\nDuplicate model for tracking duplicate photo groups.\n\nDuplicates are photos that are either:\n- EXACT_COPY: Byte-for-byte identical files (same MD5 hash, different paths)\n- VISUAL_DUPLICATE: Visually similar photos (similar perceptual hash or CLIP embeddings)\n\nDuplicates are fundamentally different from Stacks:\n- Duplicates represent redundant storage that the user may want to clean up\n- Stacks represent related photos that should be kept together for organization\n\nThis separation allows:\n- Focused workflows: Duplicates → review/delete, Stacks → browse/organize\n- Different UX: Duplicates page focused on storage savings vs Stacks for browsing\n- Clearer data model with appropriate fields for each concept\n\"\"\"\n\nimport uuid\n\nfrom django.db import models\nfrom django.utils import timezone\n\nfrom api.models.user import User, get_deleted_user\n\n\nclass Duplicate(models.Model):\n \"\"\"\n Represents a group of duplicate photos that should be reviewed.\n \n Photos in a duplicate group are candidates for deletion - the user\n reviews them and decides which to keep.\n \"\"\"\n\n class DuplicateType(models.TextChoices):\n # Exact byte-for-byte copies (same MD5 hash, different file paths)\n EXACT_COPY = \"exact_copy\", \"Exact Copies\"\n # Visually similar images (similar pHash or CLIP embeddings)\n VISUAL_DUPLICATE = \"visual_duplicate\", \"Visual Duplicates\"\n\n class ReviewStatus(models.TextChoices):\n # User hasn't reviewed yet\n PENDING = \"pending\", \"Pending Review\"\n # User selected a primary and trashed others\n RESOLVED = \"resolved\", \"Resolved\"\n # User marked as \"not actually duplicates\"\n DISMISSED = \"dismissed\", \"Dismissed\"\n\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n owner = models.ForeignKey(\n User,\n on_delete=models.SET(get_deleted_user),\n related_name=\"duplicates\",\n )\n\n duplicate_type = models.CharField(\n max_length=20,\n choices=DuplicateType.choices,\n default=DuplicateType.VISUAL_DUPLICATE,\n db_index=True,\n )\n\n review_status = models.CharField(\n max_length=20,\n choices=ReviewStatus.choices,\n default=ReviewStatus.PENDING,\n db_index=True,\n )\n\n # The photo the user chose to keep (set when resolved)\n kept_photo = models.ForeignKey(\n \"Photo\",\n on_delete=models.SET_NULL,\n null=True,\n blank=True,\n related_name=\"kept_in_duplicates\",\n )\n\n # Detection metadata\n created_at = models.DateTimeField(auto_now_add=True)\n updated_at = models.DateTimeField(auto_now=True)\n reviewed_at = models.DateTimeField(null=True, blank=True)\n\n # For visual duplicates: similarity score (0-1, higher = more similar)\n similarity_score = models.FloatField(null=True, blank=True)\n\n # Potential storage savings if non-kept photos are removed (bytes)\n potential_savings = models.BigIntegerField(default=0)\n\n # Number of photos trashed when resolved\n trashed_count = models.IntegerField(default=0)\n\n # Optional note from user\n note = models.TextField(blank=True, null=True)\n\n class Meta:\n ordering = [\"-created_at\"]\n verbose_name = \"Duplicate\"\n verbose_name_plural = \"Duplicates\"\n indexes = [\n models.Index(fields=[\"owner\", \"duplicate_type\"]),\n models.Index(fields=[\"owner\", \"review_status\"]),\n ]\n\n def __str__(self):\n return f\"Duplicate {self.id} - {self.duplicate_type} - {self.owner.username}\"\n\n @property\n def photo_count(self):\n \"\"\"Number of photos in this duplicate group.\"\"\"\n return self.photos.count()\n\n def get_photos_ordered_by_quality(self):\n \"\"\"\n Returns photos ordered by quality metrics.\n Higher resolution and larger file size are considered better quality.\n \"\"\"\n return self.photos.select_related('metadata').order_by(\n \"-metadata__width\", \"-metadata__height\", \"-size\"\n )\n\n def auto_select_best_photo(self):\n \"\"\"\n Automatically selects the best quality photo as the kept photo.\n Used as a suggestion for the user.\n \n For EXACT_COPY: Picks the one with shortest path (likely \"original\")\n For VISUAL_DUPLICATE: Highest resolution\n \n Returns:\n The best Photo instance or None\n \"\"\"\n from django.db.models.functions import Length\n \n photos = self.photos.all()\n if not photos.exists():\n return None\n\n if self.duplicate_type == self.DuplicateType.EXACT_COPY:\n # For exact copies, pick the one with shortest path (likely \"original\")\n best = photos.order_by(Length(\"main_file__path\")).first()\n else:\n # For visual duplicates: highest resolution\n from django.db.models import F\n best = photos.order_by(\n F(\"metadata__width\") * F(\"metadata__height\")\n ).last()\n\n return best\n\n def calculate_potential_savings(self):\n \"\"\"\n Calculate how much storage could be saved if non-best photos\n are removed from disk.\n \"\"\"\n best = self.auto_select_best_photo()\n if not best:\n self.potential_savings = 0\n else:\n # Sum size of all photos except best\n from django.db.models import Sum\n non_best_size = (\n self.photos.exclude(pk=best.pk)\n .aggregate(total=Sum(\"size\"))\n .get(\"total\", 0)\n ) or 0\n self.potential_savings = non_best_size\n\n self.save(update_fields=[\"potential_savings\", \"updated_at\"])\n return self.potential_savings\n\n def resolve(self, kept_photo, trash_others: bool = True):\n \"\"\"\n Mark the duplicate as resolved by selecting a photo to keep.\n \n Args:\n kept_photo: The Photo instance to keep\n trash_others: Whether to move other photos to trash\n \"\"\"\n # Set the kept photo\n self.kept_photo = kept_photo\n self.review_status = self.ReviewStatus.RESOLVED\n self.reviewed_at = timezone.now()\n\n # Trash others if requested\n if trash_others:\n other_photos = self.photos.exclude(pk=kept_photo.pk)\n self.trashed_count = other_photos.update(in_trashcan=True)\n \n self.save()\n return self\n\n def dismiss(self):\n \"\"\"Mark as 'not actually duplicates' and unlink photos from group.\"\"\"\n self.review_status = self.ReviewStatus.DISMISSED\n self.reviewed_at = timezone.now()\n \n # Unlink photos from duplicate group (ManyToMany)\n for photo in self.photos.all():\n photo.duplicates.remove(self)\n \n self.save()\n return self\n\n def revert(self):\n \"\"\"Revert a resolved duplicate, restoring trashed photos.\"\"\"\n if self.review_status != self.ReviewStatus.RESOLVED:\n return 0\n\n # Restore trashed photos in this duplicate group\n restored_count = self.photos.filter(\n in_trashcan=True\n ).update(in_trashcan=False)\n\n # Reset to pending\n self.review_status = self.ReviewStatus.PENDING\n self.kept_photo = None\n self.trashed_count = 0\n self.reviewed_at = None\n\n self.save()\n return restored_count\n\n def merge_with(self, other_duplicate: \"Duplicate\"):\n \"\"\"\n Merge another duplicate group into this one.\n All photos from the other group are moved here,\n and the other group is deleted.\n \"\"\"\n if other_duplicate.pk == self.pk:\n return\n\n # Move all photos from other duplicate to this one (ManyToMany)\n for photo in other_duplicate.photos.all():\n photo.duplicates.add(self)\n photo.duplicates.remove(other_duplicate)\n\n self.calculate_potential_savings()\n\n # Delete the now-empty duplicate group\n other_duplicate.delete()\n\n @classmethod\n def create_or_merge(cls, owner, duplicate_type, photos, similarity_score=None):\n \"\"\"\n Create a new duplicate group or merge into existing if any photo is already grouped.\n\n Args:\n owner: User who owns the photos\n duplicate_type: Type of duplicate (EXACT_COPY or VISUAL_DUPLICATE)\n photos: Queryset or list of Photo objects to group\n similarity_score: Optional similarity score for visual duplicates\n\n Returns:\n The Duplicate instance (new or existing)\n \"\"\"\n photo_list = list(photos)\n if len(photo_list) < 2:\n return None\n\n # Check if any photo is already in a duplicate group of the same type\n existing_duplicates = cls.objects.filter(\n photos__in=photo_list,\n duplicate_type=duplicate_type,\n owner=owner,\n ).distinct()\n\n if existing_duplicates.exists():\n # Merge all into the first existing duplicate group\n target_duplicate = existing_duplicates.first()\n for duplicate in existing_duplicates[1:]:\n target_duplicate.merge_with(duplicate)\n\n # Add any new photos to the group (ManyToMany)\n for photo in photo_list:\n if not photo.duplicates.filter(pk=target_duplicate.pk).exists():\n photo.duplicates.add(target_duplicate)\n\n target_duplicate.calculate_potential_savings()\n return target_duplicate\n else:\n # Create new duplicate group\n duplicate = cls.objects.create(\n owner=owner,\n duplicate_type=duplicate_type,\n similarity_score=similarity_score,\n )\n\n # Associate photos (ManyToMany)\n for photo in photo_list:\n photo.duplicates.add(duplicate)\n\n duplicate.calculate_potential_savings()\n return duplicate\n" }, { "path": "api/models/face.py", "content": "import os\n\nimport numpy as np\nfrom django.db import models\nfrom django.dispatch import receiver\n\nfrom api.face_recognition import get_face_encodings\nfrom api.models.cluster import Cluster\nfrom api.models.person import Person\nfrom api.models.photo import Photo\n\n\nclass Face(models.Model):\n photo = models.ForeignKey(\n Photo, related_name=\"faces\", on_delete=models.CASCADE, blank=False, null=True\n )\n image = models.ImageField(upload_to=\"faces\", null=True)\n\n person = models.ForeignKey(\n Person, on_delete=models.DO_NOTHING, related_name=\"faces\", null=True\n )\n\n classification_person = models.ForeignKey(\n Person,\n related_name=\"classification_faces\",\n on_delete=models.SET_NULL,\n blank=True,\n null=True,\n )\n classification_probability = models.FloatField(default=0.0, db_index=True)\n\n cluster_person = models.ForeignKey(\n Person,\n related_name=\"cluster_faces\",\n on_delete=models.SET_NULL,\n blank=True,\n null=True,\n )\n cluster_probability = models.FloatField(default=0.0, db_index=True)\n\n deleted = models.BooleanField(default=False)\n\n cluster = models.ForeignKey(\n Cluster,\n related_name=\"faces\",\n on_delete=models.SET_NULL,\n blank=True,\n null=True,\n )\n\n location_top = models.IntegerField()\n location_bottom = models.IntegerField()\n location_left = models.IntegerField()\n location_right = models.IntegerField()\n encoding = models.TextField()\n\n @property\n def timestamp(self):\n return self.photo.exif_timestamp if self.photo else None\n\n def __str__(self):\n return \"%d\" % self.id\n\n def generate_encoding(self):\n self.encoding = (\n get_face_encodings(\n self.photo.thumbnail.thumbnail_big.path,\n [\n (\n self.location_top,\n self.location_right,\n self.location_bottom,\n self.location_left,\n )\n ],\n )[0]\n .tobytes()\n .hex()\n )\n self.save()\n\n def get_encoding_array(self):\n return np.frombuffer(bytes.fromhex(self.encoding))\n\n\n@receiver(models.signals.post_delete, sender=Person)\ndef reset_person(sender, instance, **kwargs):\n instance.faces.update(person=None)\n\n\n# From: https://stackoverflow.com/questions/16041232/django-delete-filefield\n@receiver(models.signals.post_delete, sender=Face)\ndef auto_delete_file_on_delete(sender, instance, **kwargs):\n if instance.image:\n if os.path.isfile(instance.image.path):\n os.remove(instance.image.path)\n" }, { "path": "api/models/file.py", "content": "import hashlib\nimport os\n\nimport magic\nimport pyvips\nfrom django.db import models\n\nfrom api import util\n\n# Most optimal value for performance/memory. Found here:\n# https://stackoverflow.com/questions/17731660/hashlib-optimal-size-of-chunks-to-be-used-in-md5-update\nBUFFER_SIZE = 65536\n\n\n# To-Do: add owner to file\nclass File(models.Model):\n IMAGE = 1\n VIDEO = 2\n METADATA_FILE = 3\n RAW_FILE = 4\n UNKNOWN = 5\n\n FILE_TYPES = (\n (IMAGE, \"Image\"),\n (VIDEO, \"Video\"),\n (METADATA_FILE, \"Metadata File e.g. XMP\"),\n (RAW_FILE, \"Raw File\"),\n (UNKNOWN, \"Unknown\"),\n )\n\n hash = models.CharField(primary_key=True, max_length=64, null=False)\n path = models.TextField(blank=True, default=\"\", unique=True)\n type = models.PositiveIntegerField(\n blank=True,\n choices=FILE_TYPES,\n )\n missing = models.BooleanField(default=False)\n embedded_media = models.ManyToManyField(\"self\", symmetrical=False)\n\n def __str__(self):\n return self.path + \" \" + self._find_out_type()\n\n @staticmethod\n def create(path: str, user):\n \"\"\"\n Create or retrieve a File record for the given path.\n\n Uses get_or_create pattern to handle unique path constraint:\n - If a File with this path already exists, return it\n - If not, create a new File with calculated hash\n\n Handles race conditions: if concurrent creates happen for the same\n path, only one will succeed and others will return the existing file.\n\n Note: If file content has changed (different hash), the existing\n File record is returned. Hash updates should be handled separately\n during rescan operations.\n\n Args:\n path: The file system path to the file\n user: The user who owns this file (used for hash calculation)\n\n Returns:\n File: The existing or newly created File instance\n \"\"\"\n from django.db import IntegrityError\n\n # Check if a File with this path already exists\n existing = File.objects.filter(path=path).first()\n if existing:\n return existing\n\n # Create new File\n file = File()\n file.path = path\n file.hash = calculate_hash(user, path)\n file._find_out_type()\n\n try:\n file.save()\n return file\n except IntegrityError:\n # Race condition: another thread created the file between our check and save\n # Try to fetch by path first (unique constraint), then by hash (primary key)\n existing = File.objects.filter(path=path).first()\n if existing:\n return existing\n # If path doesn't exist, hash collision occurred - fetch by hash\n existing = File.objects.filter(hash=file.hash).first()\n if existing:\n return existing\n # Re-raise if we can't find the conflicting record\n raise\n\n def _find_out_type(self):\n self.type = File.IMAGE\n if is_raw(self.path):\n self.type = File.RAW_FILE\n if is_video(self.path):\n self.type = File.VIDEO\n if is_metadata(self.path):\n self.type = File.METADATA_FILE\n self.save()\n\n\ndef is_video(path):\n try:\n mime = magic.Magic(mime=True)\n filename = mime.from_file(path)\n return filename.find(\"video\") != -1\n except Exception:\n util.logger.error(f\"Error while checking if file is video: {path}\")\n return False\n\n\ndef is_raw(path):\n fileextension = os.path.splitext(path)[1]\n rawformats = [\n \".RWZ\",\n \".CR2\",\n \".NRW\",\n \".EIP\",\n \".RAF\",\n \".ERF\",\n \".RW2\",\n \".NEF\",\n \".ARW\",\n \".K25\",\n \".DNG\",\n \".SRF\",\n \".DCR\",\n \".RAW\",\n \".CRW\",\n \".BAY\",\n \".3FR\",\n \".CS1\",\n \".MEF\",\n \".ORF\",\n \".ARI\",\n \".SR2\",\n \".KDC\",\n \".MOS\",\n \".MFW\",\n \".FFF\",\n \".CR3\",\n \".SRW\",\n \".RWL\",\n \".J6I\",\n \".KC2\",\n \".X3F\",\n \".MRW\",\n \".IIQ\",\n \".PEF\",\n \".CXI\",\n \".MDC\",\n ]\n return fileextension.upper() in rawformats\n\n\ndef is_metadata(path):\n fileextension = os.path.splitext(path)[1]\n rawformats = [\n \".XMP\",\n ]\n return fileextension.upper() in rawformats\n\n\ndef is_valid_media(path, user) -> bool:\n if is_video(path=path) or is_metadata(path=path):\n return True\n if is_raw(path=path):\n return True\n try:\n pyvips.Image.thumbnail(path, 10000, height=200, size=pyvips.enums.Size.DOWN)\n return True\n except Exception as e:\n util.logger.info(f\"Could not handle {path}, because {str(e)}\")\n return False\n\n\ndef calculate_hash(user, path):\n try:\n hash_md5 = hashlib.md5()\n with open(path, \"rb\") as f:\n for chunk in iter(lambda: f.read(BUFFER_SIZE), b\"\"):\n hash_md5.update(chunk)\n return hash_md5.hexdigest() + str(user.id)\n except Exception as e:\n util.logger.error(f\"Could not calculate hash for file {path}\")\n raise e\n\n\ndef calculate_hash_b64(user, content):\n hash_md5 = hashlib.md5()\n with content as f:\n for chunk in iter(lambda: f.read(BUFFER_SIZE), b\"\"):\n hash_md5.update(chunk)\n return hash_md5.hexdigest() + str(user.id)\n" }, { "path": "api/models/long_running_job.py", "content": "import uuid\nfrom datetime import datetime, timedelta\n\nfrom django.db import models\nfrom django.utils import timezone\n\nfrom api.models.user import User, get_deleted_user\n\n\nclass LongRunningJob(models.Model):\n JOB_SCAN_PHOTOS = 1\n JOB_GENERATE_AUTO_ALBUMS = 2\n JOB_GENERATE_AUTO_ALBUM_TITLES = 3\n JOB_TRAIN_FACES = 4\n JOB_DELETE_MISSING_PHOTOS = 5\n JOB_CALCULATE_CLIP_EMBEDDINGS = 6\n JOB_SCAN_FACES = 7\n JOB_CLUSTER_ALL_FACES = 8\n JOB_DOWNLOAD_PHOTOS = 9\n JOB_DOWNLOAD_MODELS = 10\n JOB_ADD_GEOLOCATION = 11\n JOB_GENERATE_TAGS = 12\n JOB_GENERATE_FACE_EMBEDDINGS = 13\n JOB_SCAN_MISSING_PHOTOS = 14\n JOB_DETECT_DUPLICATES = 15\n JOB_REPAIR_FILE_VARIANTS = 16\n\n JOB_TYPES = (\n (JOB_SCAN_PHOTOS, \"Scan Photos\"),\n (JOB_GENERATE_AUTO_ALBUMS, \"Generate Event Albums\"),\n (JOB_GENERATE_AUTO_ALBUM_TITLES, \"Regenerate Event Titles\"),\n (JOB_TRAIN_FACES, \"Train Faces\"),\n (JOB_DELETE_MISSING_PHOTOS, \"Delete Missing Photos\"),\n (JOB_SCAN_FACES, \"Scan Faces\"),\n (JOB_CALCULATE_CLIP_EMBEDDINGS, \"Calculate Clip Embeddings\"),\n (JOB_CLUSTER_ALL_FACES, \"Find Similar Faces\"),\n (JOB_DOWNLOAD_PHOTOS, \"Download Selected Photos\"),\n (JOB_DOWNLOAD_MODELS, \"Download Models\"),\n (JOB_ADD_GEOLOCATION, \"Add Geolocation\"),\n (JOB_GENERATE_TAGS, \"Generate Tags\"),\n (JOB_GENERATE_FACE_EMBEDDINGS, \"Generate Face Embeddings\"),\n (JOB_SCAN_MISSING_PHOTOS, \"Scan Missing Photos\"),\n (JOB_DETECT_DUPLICATES, \"Detect Duplicate Photos\"),\n (JOB_REPAIR_FILE_VARIANTS, \"Repair File Variants\"),\n )\n\n job_type = models.PositiveIntegerField(\n choices=JOB_TYPES,\n )\n\n finished = models.BooleanField(default=False, blank=False, null=False)\n failed = models.BooleanField(default=False, blank=False, null=False)\n job_id = models.CharField(max_length=36, unique=True, db_index=True)\n queued_at = models.DateTimeField(default=datetime.now, null=False)\n started_at = models.DateTimeField(null=True)\n finished_at = models.DateTimeField(null=True)\n started_by = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n progress_current = models.PositiveIntegerField(default=0)\n progress_target = models.PositiveIntegerField(default=0)\n # New fields for detailed progress reporting\n progress_step = models.CharField(max_length=100, null=True, blank=True) # Current step description\n result = models.JSONField(null=True, blank=True) # Detailed result/progress data\n\n class Meta:\n ordering = [\"-queued_at\"]\n verbose_name = \"Long Running Job\"\n verbose_name_plural = \"Long Running Jobs\"\n\n def __str__(self):\n status = \"failed\" if self.failed else (\"finished\" if self.finished else \"running\" if self.started_at else \"queued\")\n return f\"Job {self.job_id} - {self.get_job_type_display()} - {status}\"\n\n @property\n def is_running(self):\n \"\"\"Check if job is currently running (started but not finished).\"\"\"\n return self.started_at is not None and not self.finished\n\n @property\n def duration(self):\n \"\"\"Return job duration in seconds, or None if not started.\"\"\"\n if not self.started_at:\n return None\n end = self.finished_at or timezone.now()\n return (end - self.started_at).total_seconds()\n\n def start(self):\n \"\"\"Mark job as started.\"\"\"\n self.started_at = timezone.now()\n self.save(update_fields=[\"started_at\"])\n\n def complete(self, result=None):\n \"\"\"Mark job as successfully completed.\"\"\"\n self.finished = True\n self.finished_at = timezone.now()\n if result is not None:\n self.result = result\n self.save(update_fields=[\"finished\", \"finished_at\", \"result\"])\n\n def fail(self, error=None):\n \"\"\"Mark job as failed with optional error message.\"\"\"\n self.failed = True\n self.finished = True\n self.finished_at = timezone.now()\n if error is not None:\n self.result = {\"status\": \"failed\", \"error\": str(error)}\n self.save(update_fields=[\"failed\", \"finished\", \"finished_at\", \"result\"])\n\n def update_progress(self, current, target=None, step=None):\n \"\"\"Update job progress counters and optional step description.\"\"\"\n update_fields = [\"progress_current\"]\n self.progress_current = current\n if target is not None:\n self.progress_target = target\n update_fields.append(\"progress_target\")\n if step is not None:\n self.progress_step = step\n update_fields.append(\"progress_step\")\n self.save(update_fields=update_fields)\n\n def set_result(self, result):\n \"\"\"Update the job result/progress data.\"\"\"\n self.result = result\n self.save(update_fields=[\"result\"])\n\n @classmethod\n def create_job(cls, user, job_type, job_id=None, start_now=False):\n \"\"\"\n Factory method to create a new job with proper defaults.\n \n Args:\n user: The user who started the job\n job_type: One of the JOB_* constants\n job_id: Optional job ID (auto-generated UUID if not provided)\n start_now: If True, set started_at to now\n \n Returns:\n The newly created LongRunningJob instance\n \"\"\"\n if job_id is None:\n job_id = str(uuid.uuid4())\n job = cls.objects.create(\n started_by=user,\n job_id=str(job_id),\n queued_at=timezone.now(),\n job_type=job_type,\n )\n if start_now:\n job.start()\n return job\n\n @classmethod\n def get_or_create_job(cls, user, job_type, job_id):\n \"\"\"\n Get an existing job by job_id or create a new one.\n \n This is useful for queued jobs where the job_id is known ahead of time.\n If the job exists, it will be marked as started. If not, a new job is created.\n \n Args:\n user: The user who started the job\n job_type: One of the JOB_* constants\n job_id: The job ID to look up or use for creation\n \n Returns:\n The LongRunningJob instance (existing or newly created)\n \"\"\"\n if cls.objects.filter(job_id=job_id).exists():\n job = cls.objects.get(job_id=job_id)\n job.start()\n return job\n return cls.create_job(user=user, job_type=job_type, job_id=job_id, start_now=True)\n\n @classmethod\n def cleanup_stuck_jobs(cls, hours=24):\n \"\"\"\n Mark jobs as failed if they've been running for too long.\n \n Jobs that have started_at set but finished=False for longer than\n the specified hours are considered stuck and will be marked as failed.\n \n Args:\n hours: Number of hours after which a running job is considered stuck\n \n Returns:\n Number of jobs marked as failed\n \"\"\"\n cutoff = timezone.now() - timedelta(hours=hours)\n stuck_jobs = cls.objects.filter(\n finished=False,\n started_at__isnull=False,\n started_at__lt=cutoff\n )\n count = stuck_jobs.count()\n stuck_jobs.update(\n failed=True,\n finished=True,\n finished_at=timezone.now(),\n result={\"status\": \"failed\", \"error\": f\"Job timed out after {hours} hours\"}\n )\n return count\n\n @classmethod\n def cleanup_old_jobs(cls, days=30):\n \"\"\"\n Delete completed/failed jobs older than specified days.\n \n Args:\n days: Number of days after which completed jobs should be deleted\n \n Returns:\n Number of jobs deleted\n \"\"\"\n cutoff = timezone.now() - timedelta(days=days)\n deleted, _ = cls.objects.filter(\n finished=True,\n finished_at__lt=cutoff\n ).delete()\n return deleted\n" }, { "path": "api/models/person.py", "content": "import datetime\n\nimport pytz\nfrom django.core.validators import MinLengthValidator\nfrom django.db import models\nfrom django.db.models import Prefetch\n\nfrom api.models.photo import Photo\nfrom api.models.user import User\n\nutc = pytz.UTC\n\n\nclass Person(models.Model):\n UNKNOWN_PERSON_NAME = \"Unknown - Other\"\n KIND_USER = \"USER\"\n KIND_CLUSTER = \"CLUSTER\"\n KIND_UNKNOWN = \"UNKNOWN\"\n KIND_CHOICES = (\n (KIND_USER, \"User Labelled\"),\n (KIND_CLUSTER, \"Cluster ID\"),\n (KIND_UNKNOWN, \"Unknown Person\"),\n )\n name = models.CharField(\n blank=False, max_length=128, validators=[MinLengthValidator(1)], db_index=True\n )\n kind = models.CharField(choices=KIND_CHOICES, max_length=10)\n cover_photo = models.ForeignKey(\n Photo, related_name=\"person\", on_delete=models.SET_NULL, blank=False, null=True\n )\n cover_face = models.ForeignKey(\n \"Face\",\n related_name=\"face\",\n on_delete=models.SET_NULL,\n blank=False,\n null=True,\n )\n face_count = models.IntegerField(default=0)\n cluster_owner = models.ForeignKey(\n User,\n related_name=\"owner\",\n on_delete=models.SET_NULL,\n default=None,\n null=True,\n )\n\n def __str__(self):\n return (\n self.name\n + \" (\"\n + self.kind\n + \")\"\n + \" (\"\n + str(self.id)\n + \")\"\n + \" (\"\n + str(self.cluster_owner)\n + \")\"\n )\n\n def _calculate_face_count(self):\n self.face_count = self.faces.filter(\n photo__hidden=False,\n photo__in_trashcan=False,\n photo__owner=self.cluster_owner.id,\n ).count()\n self.save()\n\n def _set_default_cover_photo(self):\n if not self.cover_photo and self.faces.count() > 0:\n self.cover_photo = self.faces.first().photo\n self.cover_face = self.faces.first()\n self.save()\n\n def get_photos(self, owner):\n faces = list(\n self.faces.prefetch_related(\n Prefetch(\n \"photo\",\n queryset=Photo.objects.exclude(image_hash=None)\n .filter(hidden=False, owner=owner)\n .order_by(\"-exif_timestamp\")\n .only(\n \"image_hash\",\n \"exif_timestamp\",\n \"rating\",\n \"owner__id\",\n \"public\",\n \"hidden\",\n )\n .prefetch_related(\"owner\"),\n )\n )\n )\n\n photos = [face.photo for face in faces if hasattr(face.photo, \"owner\")]\n photos.sort(\n key=lambda x: x.exif_timestamp or utc.localize(datetime.datetime.min),\n reverse=True,\n )\n return photos\n\n\n# TODO: Should be removed in the future, as it is not used, only in migrations\ndef get_unknown_person(owner: User = None):\n unknown_person: Person = Person.objects.get_or_create(\n name=Person.UNKNOWN_PERSON_NAME, cluster_owner=owner, kind=Person.KIND_UNKNOWN\n )[0]\n if unknown_person.kind != Person.KIND_UNKNOWN:\n unknown_person.kind = Person.KIND_UNKNOWN\n unknown_person.save()\n return unknown_person\n\n\ndef get_or_create_person(name, owner: User = None, kind: str = Person.KIND_UNKNOWN):\n return Person.objects.get_or_create(name=name, cluster_owner=owner, kind=kind)[0]\n" }, { "path": "api/models/photo.py", "content": "import json\nimport numbers\nimport os\nimport uuid\nfrom fractions import Fraction\nfrom io import BytesIO\n\nimport numpy as np\nimport PIL\nfrom django.core.files.base import ContentFile\nfrom django.db import models\nfrom django.db.models import Q\nfrom django.db.utils import IntegrityError\n\nimport api.models\nfrom api import date_time_extractor, face_extractor, util\nfrom api.geocode import GEOCODE_VERSION\nfrom api.geocode.geocode import reverse_geocode\nfrom api.metadata.reader import get_metadata\nfrom api.metadata.tags import Tags\nfrom api.metadata.writer import write_metadata\nfrom api.models.file import File\nfrom api.models.user import User, get_deleted_user\nfrom api.util import logger\n\n\nclass VisiblePhotoManager(models.Manager):\n def get_queryset(self):\n return (\n super()\n .get_queryset()\n .filter(\n Q(hidden=False)\n & Q(thumbnail__aspect_ratio__isnull=False)\n & Q(in_trashcan=False)\n & Q(removed=False)\n )\n )\n\n\nclass Photo(models.Model):\n # UUID primary key (like Immich) - enables flexible asset management\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n # Content hash for deduplication - unique per user\n # Format: MD5 hash + user_id (e.g., \"abc123def456...789\" + \"1\")\n image_hash = models.CharField(max_length=64, db_index=True)\n\n files = models.ManyToManyField(File)\n main_file = models.ForeignKey(\n File,\n related_name=\"main_photo\",\n on_delete=models.SET_NULL,\n blank=False,\n null=True,\n )\n\n added_on = models.DateTimeField(null=False, blank=False, db_index=True)\n\n exif_gps_lat = models.FloatField(blank=True, null=True)\n exif_gps_lon = models.FloatField(blank=True, null=True)\n exif_timestamp = models.DateTimeField(blank=True, null=True, db_index=True)\n\n exif_json = models.JSONField(blank=True, null=True)\n\n geolocation_json = models.JSONField(blank=True, null=True, db_index=True)\n\n timestamp = models.DateTimeField(blank=True, null=True, db_index=True)\n rating = models.IntegerField(default=0, db_index=True)\n in_trashcan = models.BooleanField(default=False, db_index=True)\n removed = models.BooleanField(default=False, db_index=True)\n hidden = models.BooleanField(default=False, db_index=True)\n video = models.BooleanField(default=False)\n video_length = models.TextField(blank=True, null=True)\n size = models.BigIntegerField(default=0)\n # Metadata fields (camera, lens, fstop, etc.) moved to PhotoMetadata model\n # See migration 0103_remove_photo_metadata_fields.py\n\n owner = models.ForeignKey(\n User, on_delete=models.SET(get_deleted_user), default=None\n )\n\n shared_to = models.ManyToManyField(User, related_name=\"photo_shared_to\")\n\n public = models.BooleanField(default=False, db_index=True)\n\n # Use JSONField for database compatibility (works with both PostgreSQL and SQLite)\n clip_embeddings = models.JSONField(blank=True, null=True)\n\n clip_embeddings_magnitude = models.FloatField(blank=True, null=True)\n last_modified = models.DateTimeField(auto_now=True)\n\n # Perceptual hash for duplicate detection (pHash algorithm)\n perceptual_hash = models.CharField(\n max_length=64, blank=True, null=True, db_index=True\n )\n\n # Organizational photo stacks (RAW+JPEG pairs, bursts, brackets, live photos, manual)\n # A photo can belong to multiple stacks of different types simultaneously\n stacks = models.ManyToManyField(\n \"PhotoStack\",\n blank=True,\n related_name=\"photos\",\n )\n\n # Duplicate groups (exact copies, visual duplicates)\n # Separate from stacks because duplicates are about cleanup, not organization\n duplicates = models.ManyToManyField(\n \"Duplicate\",\n blank=True,\n related_name=\"photos\",\n )\n\n # Sub-second timestamp precision for burst detection\n # Stores the fractional seconds from EXIF:SubSecTimeOriginal\n exif_timestamp_subsec = models.CharField(max_length=10, blank=True, null=True)\n\n # Camera image sequence number (for burst/sequence detection)\n # From EXIF:ImageNumber or MakerNotes\n image_sequence_number = models.IntegerField(blank=True, null=True)\n\n objects = models.Manager()\n visible = VisiblePhotoManager()\n\n _loaded_values = {}\n\n def get_clip_embeddings(self):\n \"\"\"Get clip embeddings as a list, regardless of storage format\"\"\"\n if not self.clip_embeddings:\n return None\n\n # Handle case where embeddings might be stored as JSON string\n if isinstance(self.clip_embeddings, str):\n try:\n import json\n\n return json.loads(self.clip_embeddings)\n except (json.JSONDecodeError, TypeError):\n return None\n\n return self.clip_embeddings\n\n def set_clip_embeddings(self, embeddings):\n \"\"\"Set clip embeddings, automatically handling storage format\"\"\"\n self.clip_embeddings = embeddings if embeddings else None\n\n @classmethod\n def from_db(cls, db, field_names, values):\n instance = super().from_db(db, field_names, values)\n\n # save original values, when model is loaded from database,\n # in a separate attribute on the model\n instance._loaded_values = dict(zip(field_names, values))\n\n return instance\n\n def save(\n self,\n force_insert=False,\n force_update=False,\n using=None,\n update_fields=None,\n save_metadata=True,\n ):\n modified_fields = [\n field_name\n for field_name, value in self._loaded_values.items()\n if value != getattr(self, field_name)\n ]\n user = User.objects.get(username=self.owner)\n if save_metadata and user.save_metadata_to_disk != User.SaveMetadata.OFF:\n self._save_metadata(\n modified_fields,\n user.save_metadata_to_disk == User.SaveMetadata.SIDECAR_FILE,\n )\n return super().save(\n force_insert=force_insert,\n force_update=force_update,\n using=using,\n update_fields=update_fields,\n )\n\n def _save_metadata(\n self, modified_fields=None, use_sidecar=True, metadata_types=None\n ):\n \"\"\"Write metadata tags to the photo's file or sidecar.\n\n Args:\n modified_fields: List of changed field names (from Photo.save() diff).\n When None, writes all applicable tags unconditionally.\n use_sidecar: Write to XMP sidecar file if True, media file if False.\n metadata_types: List of metadata categories to write, e.g.\n [\"ratings\", \"face_tags\"]. When None, uses default behavior\n (ratings/timestamps only, for backward compatibility).\n \"\"\"\n tags_to_write = {}\n\n write_ratings = metadata_types is None or \"ratings\" in metadata_types\n write_face_tags = metadata_types is not None and \"face_tags\" in metadata_types\n\n if write_ratings:\n if modified_fields is None or \"rating\" in modified_fields:\n tags_to_write[Tags.RATING] = self.rating\n if modified_fields is not None and \"timestamp\" in modified_fields:\n # To-Do: Only works for files and not for the sidecar file\n tags_to_write[Tags.DATE_TIME] = self.timestamp\n\n if write_face_tags:\n from api.metadata.face_regions import get_face_region_tags\n\n tags_to_write.update(get_face_region_tags(self))\n\n if tags_to_write:\n write_metadata(self.main_file.path, tags_to_write, use_sidecar=use_sidecar)\n\n def _find_album_place(self):\n return api.models.album_place.AlbumPlace.objects.filter(\n Q(photos__in=[self])\n ).all()\n\n def _find_album_date(self):\n old_album_date = None\n if self.exif_timestamp:\n possible_old_album_date = api.models.album_date.get_album_date(\n date=self.exif_timestamp.date(), owner=self.owner\n )\n if (\n possible_old_album_date is not None\n and possible_old_album_date.photos.filter(\n image_hash=self.image_hash\n ).exists()\n ):\n old_album_date = possible_old_album_date\n else:\n possible_old_album_date = api.models.album_date.get_album_date(\n date=None, owner=self.owner\n )\n if (\n possible_old_album_date is not None\n and possible_old_album_date.photos.filter(\n image_hash=self.image_hash\n ).exists()\n ):\n old_album_date = possible_old_album_date\n return old_album_date\n\n def _extract_date_time_from_exif(self, commit=True):\n def exif_getter(tags):\n return get_metadata(self.main_file.path, tags=tags, try_sidecar=True)\n\n datetime_config = json.loads(self.owner.datetime_rules)\n extracted_local_time = date_time_extractor.extract_local_date_time(\n self.main_file.path,\n date_time_extractor.as_rules(datetime_config),\n exif_getter,\n self.exif_gps_lat,\n self.exif_gps_lon,\n self.owner.default_timezone,\n self.timestamp,\n )\n\n old_album_date = self._find_album_date()\n if self.exif_timestamp != extracted_local_time:\n self.exif_timestamp = extracted_local_time\n\n if old_album_date is not None:\n old_album_date.photos.remove(self)\n old_album_date.save()\n\n album_date = None\n\n if self.exif_timestamp:\n album_date = api.models.album_date.get_or_create_album_date(\n date=self.exif_timestamp.date(), owner=self.owner\n )\n album_date.photos.add(self)\n else:\n album_date = api.models.album_date.get_or_create_album_date(\n date=None, owner=self.owner\n )\n album_date.photos.add(self)\n\n if commit:\n self.save()\n album_date.save()\n\n def _geolocate(self, commit=True):\n old_gps_lat = self.exif_gps_lat\n old_gps_lon = self.exif_gps_lon\n new_gps_lat, new_gps_lon = get_metadata(\n self.main_file.path,\n tags=[Tags.LATITUDE, Tags.LONGITUDE],\n try_sidecar=True,\n )\n old_album_places = self._find_album_place()\n # Skip if it hasn't changed or is null\n if not new_gps_lat or not new_gps_lon:\n return\n if (\n old_gps_lat == float(new_gps_lat)\n and old_gps_lon == float(new_gps_lon)\n and old_album_places.count() != 0\n and self.geolocation_json\n and \"_v\" in self.geolocation_json\n and self.geolocation_json[\"_v\"] == GEOCODE_VERSION\n ):\n return\n self.exif_gps_lon = float(new_gps_lon)\n self.exif_gps_lat = float(new_gps_lat)\n if commit:\n self.save()\n try:\n res = reverse_geocode(new_gps_lat, new_gps_lon)\n if not res:\n return\n except Exception as e:\n util.logger.warning(e)\n util.logger.warning(\"Something went wrong with geolocating\")\n return\n\n self.geolocation_json = res\n\n # Update search location through PhotoSearch model\n from api.models.photo_search import PhotoSearch\n\n search_instance, created = PhotoSearch.objects.get_or_create(photo=self)\n search_instance.update_search_location(res)\n search_instance.save()\n\n # Delete photo from album places if location has changed\n if old_album_places is not None:\n for old_album_place in old_album_places:\n old_album_place.photos.remove(self)\n old_album_place.save()\n\n # Add photo to new album places\n for geolocation_level, feature in enumerate(self.geolocation_json[\"features\"]):\n if \"text\" not in feature.keys() or feature[\"text\"].isnumeric():\n continue\n album_place = api.models.album_place.get_album_place(\n feature[\"text\"], owner=self.owner\n )\n if album_place.photos.filter(image_hash=self.image_hash).count() == 0:\n album_place.geolocation_level = (\n len(self.geolocation_json[\"features\"]) - geolocation_level\n )\n album_place.photos.add(self)\n album_place.save()\n\n if commit:\n self.save()\n\n def _add_location_to_album_dates(self):\n if not self.geolocation_json:\n return\n if len(self.geolocation_json[\"places\"]) < 2:\n logger.info(self.geolocation_json)\n return\n\n album_date = self._find_album_date()\n city_name = self.geolocation_json[\"places\"][-2]\n if album_date.location and len(album_date.location) > 0:\n prev_value = album_date.location\n new_value = prev_value\n if city_name not in prev_value[\"places\"]:\n new_value[\"places\"].append(city_name)\n new_value[\"places\"] = list(set(new_value[\"places\"]))\n album_date.location = new_value\n else:\n album_date.location = {\"places\": [city_name]}\n # Safe geolocation_json\n album_date.save()\n\n def _extract_faces(self, second_try=False):\n unknown_cluster: api.models.cluster.Cluster = (\n api.models.cluster.get_unknown_cluster(user=self.owner)\n )\n try:\n big_thumbnail_image = np.array(\n PIL.Image.open(self.thumbnail.thumbnail_big.path)\n )\n\n face_locations = face_extractor.extract(\n self.main_file.path, self.thumbnail.thumbnail_big.path, self.owner\n )\n\n if len(face_locations) == 0:\n return\n\n for idx_face, face_location in enumerate(face_locations):\n top, right, bottom, left, person_name = face_location\n if person_name:\n person = api.models.person.get_or_create_person(\n name=person_name, owner=self.owner\n )\n person.save()\n else:\n person = None\n\n face_image = big_thumbnail_image[top:bottom, left:right]\n face_image = PIL.Image.fromarray(face_image)\n\n image_path = self.image_hash + \"_\" + str(idx_face) + \".jpg\"\n\n margin = int((right - left) * 0.05)\n existing_faces = api.models.face.Face.objects.filter(\n photo=self,\n location_top__lte=top + margin,\n location_top__gte=top - margin,\n location_right__lte=right + margin,\n location_right__gte=right - margin,\n location_bottom__lte=bottom + margin,\n location_bottom__gte=bottom - margin,\n location_left__lte=left + margin,\n location_left__gte=left - margin,\n )\n\n if existing_faces.count() != 0:\n continue\n\n face = api.models.face.Face(\n photo=self,\n location_top=top,\n location_right=right,\n location_bottom=bottom,\n location_left=left,\n encoding=\"\",\n person=person,\n cluster=unknown_cluster,\n )\n if person_name:\n person._calculate_face_count()\n person._set_default_cover_photo()\n face_io = BytesIO()\n face_image.save(face_io, format=\"JPEG\")\n face.image.save(image_path, ContentFile(face_io.getvalue()))\n face_io.close()\n face.save()\n logger.info(f\"image {self.image_hash}: {len(face_locations)} face(s) saved\")\n except IntegrityError:\n # When using multiple processes, then we can save at the same time, which leads to this error\n if self.files.count() > 0:\n # print out the location of the image only if we have a path\n logger.info(f\"image {self.main_file.path}: rescan face failed\")\n if not second_try:\n self._extract_faces(True)\n elif self.files.count() > 0:\n logger.error(f\"image {self.main_file.path}: rescan face failed\")\n else:\n logger.error(f\"image {self}: rescan face failed\")\n except Exception as e:\n logger.error(f\"image {self}: scan face failed\")\n raise e\n\n def _add_to_album_thing(self):\n if (\n hasattr(self, \"caption_instance\")\n and self.caption_instance\n and self.caption_instance.captions_json\n and type(self.caption_instance.captions_json) is dict\n and \"places365\" in self.caption_instance.captions_json.keys()\n ):\n for attribute in self.caption_instance.captions_json[\"places365\"][\n \"attributes\"\n ]:\n album_thing = api.models.album_thing.get_album_thing(\n title=attribute,\n owner=self.owner,\n )\n if album_thing.photos.filter(image_hash=self.image_hash).count() == 0:\n album_thing.photos.add(self)\n album_thing.thing_type = \"places365_attribute\"\n album_thing.save()\n for category in self.caption_instance.captions_json[\"places365\"][\n \"categories\"\n ]:\n album_thing = api.models.album_thing.get_album_thing(\n title=category,\n owner=self.owner,\n )\n if album_thing.photos.filter(image_hash=self.image_hash).count() == 0:\n album_thing = api.models.album_thing.get_album_thing(\n title=category, owner=self.owner\n )\n album_thing.photos.add(self)\n album_thing.thing_type = \"places365_category\"\n album_thing.save()\n\n def _check_files(self):\n for file in self.files.all():\n if not file.path or not os.path.exists(file.path):\n self.files.remove(file)\n file.missing = True\n file.save()\n self.save()\n\n def manual_delete(self):\n # Store stack references before cleanup (ManyToMany)\n photo_stacks = list(self.stacks.all())\n\n # Store duplicate group references before cleanup (ManyToMany)\n photo_duplicates = list(self.duplicates.all())\n\n # Handle file cleanup - only delete files not shared with other Photos\n for file in self.files.all():\n # Check if this file is used by other Photos (via files M2M or as main_file)\n other_photos_using_file = (\n file.photo_set.exclude(pk=self.pk).exists()\n or file.main_photo.exclude(pk=self.pk).exists()\n )\n\n if other_photos_using_file:\n # File is shared - just unlink from this photo, don't delete\n logger.info(\n f\"File {file.path} is shared with other photos, unlinking only\"\n )\n self.files.remove(file)\n else:\n # File is only used by this photo - safe to delete\n if os.path.isfile(file.path):\n logger.info(f\"Removing photo {file.path}\")\n os.remove(file.path)\n file.delete()\n\n self.files.set([])\n self.main_file = None\n self.removed = True\n\n # Clear all stack references from this photo (ManyToMany)\n self.stacks.clear()\n\n # Clear all duplicate group references from this photo (ManyToMany)\n self.duplicates.clear()\n\n result = self.save()\n\n # Clean up stacks if they're now empty or have only one photo left\n for photo_stack in photo_stacks:\n remaining_photos = photo_stack.photos.filter(removed=False).count()\n if remaining_photos <= 1:\n # If 0 or 1 photos left, delete the stack (no longer a valid grouping)\n logger.info(\n f\"Deleting photo stack {photo_stack.id} - only {remaining_photos} photos remaining\"\n )\n # Unlink remaining photos from stack first\n for remaining_photo in photo_stack.photos.all():\n remaining_photo.stacks.remove(photo_stack)\n photo_stack.delete()\n\n # Clean up duplicate groups if they're now empty or have only one photo left\n for duplicate in photo_duplicates:\n remaining_photos = duplicate.photos.filter(removed=False).count()\n if remaining_photos <= 1:\n # If 0 or 1 photos left, delete the duplicate group (no longer valid)\n logger.info(\n f\"Deleting duplicate group {duplicate.id} - only {remaining_photos} photos remaining\"\n )\n # Unlink remaining photos from duplicate first\n for remaining_photo in duplicate.photos.all():\n remaining_photo.duplicates.remove(duplicate)\n duplicate.delete()\n\n # To-Do: Handle wrong file permissions\n return result\n\n def _set_embedded_media(self, obj):\n return obj.main_file.embedded_media\n\n def __str__(self):\n main_file_path = (\n self.main_file.path if self.main_file is not None else \"No main file\"\n )\n return f\"{self.image_hash} - {self.owner} - {main_file_path}\"\n" }, { "path": "api/models/photo_caption.py", "content": "from django.db import models\nfrom django.db.models import Q\n\nimport api.models\nfrom api import util\nfrom api.image_captioning import generate_caption\nfrom api.llm import generate_prompt\nfrom api.models.user import User\n\n\nclass PhotoCaption(models.Model):\n \"\"\"Model for handling image captions and related functionality\"\"\"\n\n photo = models.OneToOneField(\n \"Photo\",\n on_delete=models.CASCADE,\n related_name=\"caption_instance\",\n primary_key=True,\n )\n captions_json = models.JSONField(blank=True, null=True, db_index=True)\n\n created_at = models.DateTimeField(auto_now_add=True)\n updated_at = models.DateTimeField(auto_now=True)\n\n class Meta:\n db_table = \"api_photo_caption\"\n\n def __str__(self):\n return f\"Captions for {self.photo.image_hash}\"\n\n def generate_captions_im2txt(self, commit=True):\n \"\"\"Generate im2txt captions for the photo\"\"\"\n if not self.photo.thumbnail or not self.photo.thumbnail.thumbnail_big:\n util.logger.warning(\n f\"No thumbnail available for photo {self.photo.image_hash}\"\n )\n return False\n\n util.logger.info(\"Generating captions with Im2txt\")\n\n try:\n image_path = self.photo.thumbnail.thumbnail_big.path\n except Exception:\n util.logger.warning(\n f\"Cannot access thumbnail path for photo {self.photo.image_hash}\"\n )\n return False\n if self.captions_json is None:\n self.captions_json = {}\n captions = self.captions_json\n\n try:\n from constance import config as site_config\n\n if site_config.CAPTIONING_MODEL == \"None\":\n util.logger.info(\"Generating captions is disabled\")\n return False\n\n if site_config.CAPTIONING_MODEL == \"moondream\":\n util.logger.info(\"Generating captions with Moondream\")\n return self._generate_captions_moondream(commit=commit)\n\n blip = False\n if site_config.CAPTIONING_MODEL == \"blip_base_capfilt_large\":\n blip = True\n\n caption = generate_caption(image_path=image_path, blip=blip)\n caption = caption.replace(\"\", \"\").replace(\"\", \"\").strip()\n\n settings = User.objects.get(username=self.photo.owner).llm_settings\n if site_config.LLM_MODEL != \"None\" and settings[\"enabled\"]:\n face = api.models.Face.objects.filter(photo=self.photo).first()\n person_name = \"\"\n if face and settings[\"add_person\"]:\n person_name = \" Person: \" + face.person.name\n place = \"\"\n if (\n self.photo.search_instance\n and self.photo.search_instance.search_location\n and settings[\"add_location\"]\n ):\n place = \" Place: \" + self.photo.search_instance.search_location\n keywords = \"\"\n if settings[\"add_keywords\"]:\n keywords = \" and tags or keywords\"\n prompt = (\n \"Q: Your task is to improve the following image caption: \"\n + caption\n + \". You also know the following information about the image:\"\n + place\n + person_name\n + \". Stick as closely as possible to the caption, while replacing generic information with information you know about the image. Only output the caption\"\n + keywords\n + \". \\n A:\"\n )\n util.logger.info(prompt)\n caption = generate_prompt(prompt)\n\n captions[\"im2txt\"] = caption\n self.captions_json = captions\n self.recreate_search_captions()\n if commit:\n self.save()\n\n util.logger.info(\n f\"generated im2txt captions for image {image_path} with SiteConfig {site_config.CAPTIONING_MODEL} with Blip: {blip} caption: {caption}\"\n )\n return True\n except Exception:\n util.logger.exception(\n f\"could not generate im2txt captions for image {image_path}\"\n )\n return False\n\n def _generate_captions_moondream(self, commit=True):\n \"\"\"Generate captions using Moondream with enhanced prompt\"\"\"\n if not self.photo.thumbnail or not self.photo.thumbnail.thumbnail_big:\n util.logger.warning(\n f\"No thumbnail available for photo {self.photo.image_hash}\"\n )\n return False\n\n try:\n image_path = self.photo.thumbnail.thumbnail_big.path\n except Exception:\n util.logger.warning(\n f\"Cannot access thumbnail path for photo {self.photo.image_hash}\"\n )\n return False\n if self.captions_json is None:\n self.captions_json = {}\n captions = self.captions_json\n\n try:\n from constance import config as site_config\n\n util.logger.info(\"Generating Moondream captions\")\n\n settings = User.objects.get(username=self.photo.owner).llm_settings\n\n # Default prompt\n prompt = \"Describe this image in a short, natural image caption.\"\n\n # Enhanced prompting if LLM is enabled\n if site_config.LLM_MODEL != \"None\" and settings[\"enabled\"]:\n face = api.models.Face.objects.filter(photo=self.photo).first()\n person_name = \"\"\n if face and settings[\"add_person\"]:\n person_name = (\n f\" The person in the photo is named {face.person.name}. \"\n f\"Use the name '{face.person.name}' directly in the caption — do not say 'a person named'. \"\n f\"Keep the caption casual and to the point, like a friend tagging a photo.\"\n )\n\n place = \"\"\n if (\n self.photo.search_instance\n and self.photo.search_instance.search_location\n and settings[\"add_location\"]\n ):\n place = f\" This photo was taken at {self.photo.search_instance.search_location}.\"\n\n keywords_instruction = \"\"\n if settings[\"add_keywords\"]:\n keywords_instruction = \" Include relevant tags and keywords.\"\n\n prompt = (\n \"Write a short, natural image caption.\"\n + person_name\n + place\n + keywords_instruction\n )\n\n util.logger.info(f\"Moondream prompt: {prompt}\")\n\n # Generate caption with the final prompt\n caption = generate_prompt(image_path=image_path, prompt=prompt)\n caption = caption.replace(\"\", \"\").replace(\"\", \"\").strip()\n\n # Save the result\n captions[\"im2txt\"] = caption\n self.captions_json = captions\n self.recreate_search_captions()\n if commit:\n self.save()\n\n util.logger.info(\n f\"Generated Moondream captions for image {image_path}, caption: {caption}\"\n )\n return True\n except Exception:\n util.logger.exception(\n f\"Could not generate Moondream captions for image {image_path}\"\n )\n return False\n\n def save_user_caption(self, caption, commit=True):\n \"\"\"Save user-provided caption\"\"\"\n if not self.photo.thumbnail or not self.photo.thumbnail.thumbnail_big:\n util.logger.warning(\n f\"No thumbnail available for photo {self.photo.image_hash}\"\n )\n return False\n\n try:\n image_path = self.photo.thumbnail.thumbnail_big.path\n except Exception:\n util.logger.warning(\n f\"Cannot access thumbnail path for photo {self.photo.image_hash}\"\n )\n return False\n\n try:\n caption = caption.replace(\"\", \"\").replace(\"\", \"\").strip()\n\n if self.captions_json is None:\n self.captions_json = {}\n self.captions_json[\"user_caption\"] = caption\n self.recreate_search_captions()\n\n if commit:\n self.save()\n\n util.logger.info(\n f\"saved captions for image {image_path}. caption: {caption}. captions_json: {self.captions_json}.\"\n )\n\n # Handle hashtags\n hashtags = [\n word\n for word in caption.split()\n if word.startswith(\"#\") and len(word) > 1\n ]\n\n for hashtag in hashtags:\n album_thing = api.models.album_thing.get_album_thing(\n title=hashtag,\n owner=self.photo.owner,\n thing_type=\"hashtag_attribute\",\n )\n if (\n album_thing.photos.filter(image_hash=self.photo.image_hash).count()\n == 0\n ):\n album_thing.photos.add(self.photo)\n album_thing.save()\n\n for album_thing in api.models.album_thing.AlbumThing.objects.filter(\n Q(photos__in=[self.photo])\n & Q(thing_type=\"hashtag_attribute\")\n & Q(owner=self.photo.owner)\n ).all():\n if album_thing.title not in caption:\n album_thing.photos.remove(self.photo)\n album_thing.save()\n return True\n except Exception:\n util.logger.exception(f\"could not save captions for image {image_path}\")\n return False\n\n def recreate_search_captions(self):\n \"\"\"Recreate search captions - directly access PhotoSearch model\"\"\"\n from api.models.photo_search import PhotoSearch\n\n search_instance, created = PhotoSearch.objects.get_or_create(photo=self.photo)\n search_instance.recreate_search_captions()\n search_instance.save()\n\n def generate_tag_captions(self, commit=True):\n \"\"\"Generate tag captions using the active tagging model (Places365 or SigLIP 2).\n\n Tags are stored per-model in captions_json and are never deleted when\n switching models -- only the active model's tags are generated / visible.\n \"\"\"\n from constance import config as site_config\n\n tagging_model = site_config.TAGGING_MODEL\n\n if not self.photo.thumbnail or not self.photo.thumbnail.thumbnail_big:\n return\n\n # Skip if this photo already has tags from the active model\n if (\n self.captions_json is not None\n and self.captions_json.get(tagging_model) is not None\n ):\n return\n\n try:\n import requests\n\n image_path = self.photo.thumbnail.thumbnail_big.path\n confidence = self.photo.owner.confidence\n json_data = {\n \"image_path\": image_path,\n \"confidence\": confidence,\n \"tagging_model\": tagging_model,\n }\n response = requests.post(\n \"http://localhost:8011/generate-tags\", json=json_data\n )\n tags_result = response.json()[\"tags\"]\n\n if tags_result is None:\n return\n if self.captions_json is None:\n self.captions_json = {}\n\n # Store under the model-specific key\n self.captions_json[tagging_model] = tags_result\n self.recreate_search_captions()\n\n if tagging_model == \"siglip2\":\n self._update_siglip2_album_things(tags_result)\n else:\n self._update_places365_album_things(tags_result)\n\n if commit:\n self.save()\n util.logger.info(\n f\"generated {tagging_model} tags for image {image_path}.\"\n )\n except Exception as e:\n util.logger.exception(\n f\"could not generate tags for image \"\n f\"{self.photo.main_file.path if self.photo.main_file else 'no main file'}\"\n )\n raise e\n\n def _update_places365_album_things(self, res_places365):\n \"\"\"Create/update AlbumThing entries for Places365 tags.\"\"\"\n # Remove old album associations for this photo\n for album_thing in api.models.album_thing.AlbumThing.objects.filter(\n Q(photos__in=[self.photo])\n & (\n Q(thing_type=\"places365_attribute\")\n | Q(thing_type=\"places365_category\")\n )\n & Q(owner=self.photo.owner)\n ).all():\n album_thing.photos.remove(self.photo)\n album_thing.save()\n\n if \"attributes\" in res_places365:\n for attribute in res_places365[\"attributes\"]:\n album_thing = api.models.album_thing.get_album_thing(\n title=attribute,\n owner=self.photo.owner,\n thing_type=\"places365_attribute\",\n )\n album_thing.photos.add(self.photo)\n album_thing.save()\n\n if \"categories\" in res_places365:\n for category in res_places365[\"categories\"]:\n album_thing = api.models.album_thing.get_album_thing(\n title=category,\n owner=self.photo.owner,\n thing_type=\"places365_category\",\n )\n album_thing.photos.add(self.photo)\n album_thing.save()\n\n def _update_siglip2_album_things(self, siglip2_result):\n \"\"\"Create/update AlbumThing entries for SigLIP 2 tags.\"\"\"\n tags = siglip2_result.get(\"tags\", [])\n\n # Remove old siglip2 album associations for this photo\n for album_thing in api.models.album_thing.AlbumThing.objects.filter(\n Q(photos__in=[self.photo])\n & Q(thing_type=\"siglip2_tag\")\n & Q(owner=self.photo.owner)\n ).all():\n album_thing.photos.remove(self.photo)\n album_thing.save()\n\n for tag in tags:\n album_thing = api.models.album_thing.get_album_thing(\n title=tag,\n owner=self.photo.owner,\n thing_type=\"siglip2_tag\",\n )\n album_thing.photos.add(self.photo)\n album_thing.save()\n\n # Backward-compatible alias\n def generate_places365_captions(self, commit=True):\n return self.generate_tag_captions(commit=commit)\n" }, { "path": "api/models/photo_metadata.py", "content": "\"\"\"\nPhotoMetadata and MetadataFile models for structured metadata handling.\n\nThis module provides proper models for:\n1. PhotoMetadata - Structured EXIF/metadata extracted from photos\n2. MetadataFile - XMP sidecars and other metadata files\n\nBenefits over current approach:\n- Typed fields instead of JSON blob (exif_json)\n- Proper foreign key relationships\n- Versioning for metadata changes\n- Support for multiple metadata sources (embedded, sidecar, user-edited)\n- Clear separation between camera metadata and derived metadata\n\"\"\"\n\nimport numbers\nimport uuid\nfrom fractions import Fraction\n\nfrom django.db import models\n\nfrom api.metadata.reader import get_metadata\nfrom api.metadata.tags import Tags\nfrom api.models.user import User, get_deleted_user\n\n\nclass MetadataFile(models.Model):\n \"\"\"\n Represents a metadata sidecar file (XMP, JSON, etc.) associated with a photo.\n\n A photo can have multiple metadata files:\n - XMP sidecar from camera/software\n - Adobe Lightroom .xmp\n - Darktable .xmp\n - User-created metadata file\n \"\"\"\n\n class FileType(models.TextChoices):\n XMP = \"xmp\", \"XMP Sidecar\"\n JSON = \"json\", \"JSON Metadata\"\n EXIF = \"exif\", \"EXIF Extract\"\n OTHER = \"other\", \"Other\"\n\n class Source(models.TextChoices):\n # Sidecar that came with the photo (same directory)\n ORIGINAL = \"original\", \"Original Sidecar\"\n # Generated by photo editing software\n SOFTWARE = \"software\", \"Software Generated\"\n # Created by LibrePhotos\n LIBREPHOTOS = \"librephotos\", \"LibrePhotos Generated\"\n # User manually created/uploaded\n USER = \"user\", \"User Created\"\n\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n # Link to the File model for actual file storage\n file = models.OneToOneField(\n \"File\",\n on_delete=models.CASCADE,\n related_name=\"metadata_info\",\n )\n\n # The photo this metadata belongs to\n photo = models.ForeignKey(\n \"Photo\",\n on_delete=models.CASCADE,\n related_name=\"metadata_files\",\n )\n\n file_type = models.CharField(\n max_length=10,\n choices=FileType.choices,\n default=FileType.XMP,\n )\n\n source = models.CharField(\n max_length=20,\n choices=Source.choices,\n default=Source.ORIGINAL,\n )\n\n # Priority for metadata resolution (higher = more authoritative)\n # User edits > Software sidecars > Original sidecars > Embedded\n priority = models.IntegerField(default=0)\n\n # Software that created this file (if known)\n creator_software = models.CharField(max_length=100, blank=True, null=True)\n\n created_at = models.DateTimeField(auto_now_add=True)\n updated_at = models.DateTimeField(auto_now=True)\n\n class Meta:\n ordering = [\"-priority\", \"-updated_at\"]\n verbose_name = \"Metadata File\"\n verbose_name_plural = \"Metadata Files\"\n\n def __str__(self):\n return f\"{self.file_type} sidecar for {self.photo_id}\"\n\n\nclass PhotoMetadata(models.Model):\n \"\"\"\n Structured metadata for a photo.\n\n This model stores extracted and computed metadata in typed fields,\n making it queryable and more maintainable than a JSON blob.\n\n Metadata Sources (in priority order):\n 1. User edits (highest priority)\n 2. XMP sidecar files\n 3. Embedded EXIF/IPTC/XMP\n 4. Computed/derived values (lowest priority)\n \"\"\"\n\n class Source(models.TextChoices):\n # Embedded in the file (EXIF, IPTC, embedded XMP)\n EMBEDDED = \"embedded\", \"Embedded in File\"\n # From XMP sidecar file\n SIDECAR = \"sidecar\", \"XMP Sidecar\"\n # User manually edited\n USER_EDIT = \"user_edit\", \"User Edit\"\n # Computed by LibrePhotos (AI, geocoding, etc.)\n COMPUTED = \"computed\", \"Computed\"\n\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n photo = models.OneToOneField(\n \"Photo\",\n on_delete=models.CASCADE,\n related_name=\"metadata\",\n )\n\n # ==================== Camera Settings ====================\n # Aperture (f-stop), e.g., 2.8\n aperture = models.FloatField(null=True, blank=True, db_index=True)\n\n # Shutter speed as a string fraction, e.g., \"1/250\"\n shutter_speed = models.CharField(max_length=20, null=True, blank=True)\n\n # Shutter speed in seconds (for queries), e.g., 0.004\n shutter_speed_seconds = models.FloatField(null=True, blank=True)\n\n # ISO sensitivity\n iso = models.IntegerField(null=True, blank=True, db_index=True)\n\n # Focal length in mm\n focal_length = models.FloatField(null=True, blank=True)\n\n # 35mm equivalent focal length\n focal_length_35mm = models.IntegerField(null=True, blank=True)\n\n # Exposure compensation in EV\n exposure_compensation = models.FloatField(null=True, blank=True)\n\n # Flash fired\n flash_fired = models.BooleanField(null=True, blank=True)\n\n # Metering mode\n metering_mode = models.CharField(max_length=50, null=True, blank=True)\n\n # White balance\n white_balance = models.CharField(max_length=50, null=True, blank=True)\n\n # ==================== Camera/Lens Info ====================\n camera_make = models.CharField(max_length=100, null=True, blank=True, db_index=True)\n camera_model = models.CharField(\n max_length=100, null=True, blank=True, db_index=True\n )\n lens_make = models.CharField(max_length=100, null=True, blank=True)\n lens_model = models.CharField(max_length=100, null=True, blank=True, db_index=True)\n serial_number = models.CharField(max_length=100, null=True, blank=True)\n\n # ==================== Image Properties ====================\n width = models.IntegerField(null=True, blank=True)\n height = models.IntegerField(null=True, blank=True)\n orientation = models.IntegerField(null=True, blank=True)\n color_space = models.CharField(max_length=50, null=True, blank=True)\n bit_depth = models.IntegerField(null=True, blank=True)\n\n # ==================== Timestamps ====================\n # Original capture time from EXIF\n date_taken = models.DateTimeField(null=True, blank=True, db_index=True)\n\n # Sub-second precision (from EXIF:SubSecTimeOriginal)\n date_taken_subsec = models.CharField(max_length=10, null=True, blank=True)\n\n # When the file was last modified\n date_modified = models.DateTimeField(null=True, blank=True)\n\n # Timezone offset if available\n timezone_offset = models.CharField(max_length=10, null=True, blank=True)\n\n # ==================== Location ====================\n gps_latitude = models.FloatField(null=True, blank=True, db_index=True)\n gps_longitude = models.FloatField(null=True, blank=True, db_index=True)\n gps_altitude = models.FloatField(null=True, blank=True)\n\n # Location from reverse geocoding\n location_country = models.CharField(\n max_length=100, null=True, blank=True, db_index=True\n )\n location_state = models.CharField(max_length=100, null=True, blank=True)\n location_city = models.CharField(\n max_length=100, null=True, blank=True, db_index=True\n )\n location_address = models.TextField(null=True, blank=True)\n\n # ==================== Content Description ====================\n title = models.CharField(max_length=500, null=True, blank=True)\n caption = models.TextField(null=True, blank=True)\n keywords = models.JSONField(null=True, blank=True) # List of strings\n\n # Rating (0-5 stars)\n rating = models.IntegerField(null=True, blank=True, db_index=True)\n\n # Copyright info\n copyright = models.TextField(null=True, blank=True)\n creator = models.CharField(max_length=200, null=True, blank=True)\n\n # ==================== Tracking ====================\n # Which source this metadata came from\n source = models.CharField(\n max_length=20,\n choices=Source.choices,\n default=Source.EMBEDDED,\n )\n\n # Raw EXIF/XMP data for fields we don't have explicit columns for\n raw_exif = models.JSONField(null=True, blank=True)\n raw_xmp = models.JSONField(null=True, blank=True)\n raw_iptc = models.JSONField(null=True, blank=True)\n\n # Version tracking\n version = models.IntegerField(default=1)\n created_at = models.DateTimeField(auto_now_add=True)\n updated_at = models.DateTimeField(auto_now=True)\n\n class Meta:\n verbose_name = \"Photo Metadata\"\n verbose_name_plural = \"Photo Metadata\"\n indexes = [\n models.Index(fields=[\"camera_make\", \"camera_model\"]),\n models.Index(fields=[\"date_taken\"]),\n models.Index(fields=[\"location_country\", \"location_city\"]),\n ]\n\n def __str__(self):\n return f\"Metadata for {self.photo_id}\"\n\n @property\n def resolution(self):\n \"\"\"Returns resolution as 'WxH' string.\"\"\"\n if self.width and self.height:\n return f\"{self.width}x{self.height}\"\n return None\n\n @property\n def megapixels(self):\n \"\"\"Returns megapixel count.\"\"\"\n if self.width and self.height:\n return round((self.width * self.height) / 1_000_000, 1)\n return None\n\n @property\n def has_location(self):\n \"\"\"Check if GPS coordinates are available.\"\"\"\n return self.gps_latitude is not None and self.gps_longitude is not None\n\n @property\n def camera_display(self):\n \"\"\"Returns a human-readable camera name.\"\"\"\n if self.camera_make and self.camera_model:\n # Avoid duplicating make in model\n if self.camera_model.startswith(self.camera_make):\n return self.camera_model\n return f\"{self.camera_make} {self.camera_model}\"\n return self.camera_model or self.camera_make\n\n @property\n def lens_display(self):\n \"\"\"Returns a human-readable lens name.\"\"\"\n if self.lens_make and self.lens_model:\n if self.lens_model.startswith(self.lens_make):\n return self.lens_model\n return f\"{self.lens_make} {self.lens_model}\"\n return self.lens_model or self.lens_make\n\n @classmethod\n def extract_exif_data(cls, photo, commit=True):\n \"\"\"\n Extract EXIF data from a photo's main file and update PhotoMetadata.\n\n This method extracts metadata from the photo's main file and:\n 1. Updates Photo fields (size, video_length, rating, exif_timestamp_subsec, image_sequence_number)\n 2. Gets or creates PhotoMetadata and populates it with extracted data\n\n Args:\n photo: Photo instance to extract metadata from\n commit: Whether to save Photo and PhotoMetadata after extraction\n\n Returns:\n PhotoMetadata instance\n \"\"\"\n if not photo.main_file:\n return None\n\n (\n size,\n fstop,\n focal_length,\n iso,\n shutter_speed,\n camera,\n lens,\n width,\n height,\n focalLength35Equivalent,\n subjectDistance,\n digitalZoomRatio,\n video_length,\n rating,\n subsec_time_original,\n image_number,\n ) = get_metadata( # noqa: E501\n photo.main_file.path,\n tags=[\n Tags.FILE_SIZE,\n Tags.FSTOP,\n Tags.FOCAL_LENGTH,\n Tags.ISO,\n Tags.EXPOSURE_TIME,\n Tags.CAMERA,\n Tags.LENS,\n Tags.IMAGE_WIDTH,\n Tags.IMAGE_HEIGHT,\n Tags.FOCAL_LENGTH_35MM,\n Tags.SUBJECT_DISTANCE,\n Tags.DIGITAL_ZOOM_RATIO,\n Tags.QUICKTIME_DURATION,\n Tags.RATING,\n Tags.SUBSEC_TIME_ORIGINAL,\n Tags.IMAGE_NUMBER,\n ],\n try_sidecar=True,\n )\n\n # Fields still on Photo model\n if size and isinstance(size, numbers.Number):\n photo.size = size\n if video_length and isinstance(video_length, numbers.Number):\n photo.video_length = video_length\n if rating and isinstance(rating, numbers.Number):\n photo.rating = rating\n\n # Burst/sequence detection fields\n if subsec_time_original:\n # SubSecTimeOriginal is typically a string like \"123\" representing milliseconds\n photo.exif_timestamp_subsec = str(subsec_time_original)[:10]\n if image_number and isinstance(image_number, numbers.Number):\n photo.image_sequence_number = int(image_number)\n\n if commit:\n photo.save()\n\n # Store metadata in PhotoMetadata model\n metadata, created = cls.objects.get_or_create(\n photo=photo, defaults={\"source\": cls.Source.EMBEDDED}\n )\n\n if fstop and isinstance(fstop, numbers.Number):\n metadata.aperture = fstop\n if focal_length and isinstance(focal_length, numbers.Number):\n metadata.focal_length = focal_length\n if iso and isinstance(iso, numbers.Number):\n metadata.iso = iso\n if shutter_speed and isinstance(shutter_speed, numbers.Number):\n metadata.shutter_speed = str(\n Fraction(shutter_speed).limit_denominator(1000)\n )\n if camera and isinstance(camera, str):\n metadata.camera_model = camera\n if lens and isinstance(lens, str):\n metadata.lens_model = lens\n if width and isinstance(width, numbers.Number):\n metadata.width = width\n if height and isinstance(height, numbers.Number):\n metadata.height = height\n if focalLength35Equivalent and isinstance(\n focalLength35Equivalent, numbers.Number\n ):\n metadata.focal_length_35mm = focalLength35Equivalent\n if rating and isinstance(rating, numbers.Number):\n metadata.rating = rating\n if subsec_time_original:\n metadata.date_taken_subsec = str(subsec_time_original)[:10]\n\n if commit:\n metadata.save()\n\n return metadata\n\n\nclass MetadataEdit(models.Model):\n \"\"\"\n Tracks user edits to photo metadata for history/undo functionality.\n\n Each edit records what field was changed, old value, new value,\n and who made the change.\n \"\"\"\n\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n photo = models.ForeignKey(\n \"Photo\",\n on_delete=models.CASCADE,\n related_name=\"metadata_edits\",\n )\n\n user = models.ForeignKey(\n User,\n on_delete=models.SET(get_deleted_user),\n related_name=\"metadata_edits\",\n )\n\n # Which field was edited\n field_name = models.CharField(max_length=100)\n\n # JSON-serialized old and new values\n old_value = models.JSONField(null=True, blank=True)\n new_value = models.JSONField(null=True, blank=True)\n\n # Whether this edit has been written back to the file\n synced_to_file = models.BooleanField(default=False)\n synced_at = models.DateTimeField(null=True, blank=True)\n\n created_at = models.DateTimeField(auto_now_add=True)\n\n class Meta:\n ordering = [\"-created_at\"]\n verbose_name = \"Metadata Edit\"\n verbose_name_plural = \"Metadata Edits\"\n indexes = [\n models.Index(fields=[\"photo\", \"-created_at\"]),\n ]\n\n def __str__(self):\n return f\"Edit {self.field_name} on {self.photo_id}\"\n" }, { "path": "api/models/photo_search.py", "content": "from django.db import models\n\nimport api.models\nfrom api import util\n\n\nclass PhotoSearch(models.Model):\n \"\"\"Model for handling photo search functionality\"\"\"\n\n photo = models.OneToOneField(\n \"Photo\",\n on_delete=models.CASCADE,\n related_name=\"search_instance\",\n primary_key=True,\n )\n search_captions = models.TextField(blank=True, null=True, db_index=True)\n search_location = models.TextField(blank=True, null=True, db_index=True)\n\n created_at = models.DateTimeField(auto_now_add=True)\n updated_at = models.DateTimeField(auto_now=True)\n\n class Meta:\n db_table = \"api_photo_search\"\n\n def __str__(self):\n return f\"Search data for {self.photo.image_hash}\"\n\n def recreate_search_captions(self):\n \"\"\"Recreate search captions from all caption sources.\n\n Only tags from the active TAGGING_MODEL are indexed into search_captions.\n This allows instant switching of tag visibility without re-inference.\n \"\"\"\n from constance import config as site_config\n\n search_captions = \"\"\n\n # Get captions from the PhotoCaption model\n if hasattr(self.photo, \"caption_instance\") and self.photo.caption_instance:\n captions_json = self.photo.caption_instance.captions_json\n if captions_json:\n # Index tags from the active tagging model only\n tagging_model = site_config.TAGGING_MODEL\n\n if tagging_model == \"siglip2\":\n siglip2_data = captions_json.get(\"siglip2\", {})\n siglip2_tags = siglip2_data.get(\"tags\", [])\n if siglip2_tags:\n search_captions += \" \".join(siglip2_tags) + \" \"\n else:\n places365_captions = captions_json.get(\"places365\", {})\n attributes = places365_captions.get(\"attributes\", [])\n search_captions += \" \".join(attributes) + \" \"\n categories = places365_captions.get(\"categories\", [])\n search_captions += \" \".join(categories) + \" \"\n environment = places365_captions.get(\"environment\", \"\")\n search_captions += environment + \" \"\n\n user_caption = captions_json.get(\"user_caption\", \"\")\n if user_caption:\n search_captions += user_caption + \" \"\n\n im2txt_caption = captions_json.get(\"im2txt\", \"\")\n if im2txt_caption:\n search_captions += im2txt_caption + \" \"\n\n # Add face/person names\n for face in api.models.face.Face.objects.filter(photo=self.photo).all():\n if face.person:\n search_captions += face.person.name + \" \"\n\n # Add file paths\n if self.photo.main_file:\n search_captions += self.photo.main_file.path + \" \"\n for file in self.photo.files.all():\n search_captions += file.path + \" \"\n\n # Add media type\n if self.photo.video:\n search_captions += \"type: video \"\n\n # Add camera and lens info from PhotoMetadata\n try:\n metadata = self.photo.metadata\n if metadata.camera_display:\n search_captions += metadata.camera_display + \" \"\n if metadata.lens_display:\n search_captions += metadata.lens_display + \" \"\n except Exception:\n # PhotoMetadata may not exist yet\n pass\n\n self.search_captions = search_captions.strip()\n\n util.logger.debug(\n f\"Recreated search captions for image {self.photo.image_hash}.\"\n )\n\n def update_search_location(self, geolocation_json):\n \"\"\"Update search location from geolocation data\"\"\"\n if geolocation_json and \"address\" in geolocation_json:\n self.search_location = geolocation_json[\"address\"]\n elif geolocation_json and \"features\" in geolocation_json:\n # Handle features format used in tests\n features = geolocation_json[\"features\"]\n location_parts = [\n feature.get(\"text\", \"\") for feature in features if feature.get(\"text\")\n ]\n self.search_location = \", \".join(location_parts) if location_parts else \"\"\n else:\n self.search_location = \"\"\n\n util.logger.debug(\n f\"Updated search location for image {self.photo.image_hash}: {self.search_location}\"\n )\n" }, { "path": "api/models/photo_stack.py", "content": "\"\"\"\nPhotoStack model for organizational photo grouping.\n\nStacks represent related photos that should be kept together for organization:\n- Burst sequences (rapid succession shots)\n- Exposure brackets (HDR sequences)\n- Manual user groupings\n\nNOTE: RAW+JPEG pairs and Live Photos are NO LONGER handled as stacks.\nInstead, they use the Photo.files ManyToMany field to store multiple file\nvariants of the same capture (PhotoPrism-like model). This allows:\n- A single Photo entity for RAW+JPEG (same capture, different formats)\n- A single Photo entity for Live Photos (image + video variant)\n- Photo stacks for DIFFERENT captures that are logically related\n\nLegacy RAW_JPEG_PAIR and LIVE_PHOTO stack types are kept for migration\ncompatibility but are deprecated and will be converted to file variants.\n\nInspired by PhotoPrism's file variant model and Immich's stacking system.\n\"\"\"\n\nimport uuid\n\nfrom django.db import models\n\nfrom api.models.user import User, get_deleted_user\n\n\nclass PhotoStack(models.Model):\n \"\"\"\n Represents a group of related but DISTINCT photos that should be kept together.\n Only the primary photo is shown in the timeline, with others accessible via expansion.\n \n NOTE: This is for grouping DIFFERENT captures (bursts, brackets, manual).\n For same-capture file variants (RAW+JPEG, Live Photos), use Photo.files instead.\n \n Stacks are informational groupings - they help organize related photos\n but don't imply that any should be deleted (unlike Duplicates).\n \"\"\"\n\n class StackType(models.TextChoices):\n # === ACTIVE STACK TYPES (for different captures) ===\n \n # Photos taken in rapid succession (burst/continuous mode)\n # User may want to browse all or pick the best\n BURST_SEQUENCE = \"burst\", \"Burst Sequence\"\n # Exposure bracketed shots (for HDR)\n # HDR processing may need all exposures\n EXPOSURE_BRACKET = \"bracket\", \"Exposure Bracket\"\n # User manually grouped photos\n # User explicitly created the grouping\n MANUAL = \"manual\", \"Manual Stack\"\n \n # === DEPRECATED STACK TYPES (migrated to Photo.files) ===\n # Kept for backwards compatibility during migration\n \n # DEPRECATED: Use Photo.files for RAW+JPEG variants\n RAW_JPEG_PAIR = \"raw_jpeg\", \"RAW + JPEG Pair (Deprecated)\"\n # DEPRECATED: Use Photo.files for Live Photo variants\n LIVE_PHOTO = \"live_photo\", \"Live Photo (Deprecated)\"\n\n # Valid stack types for new stacks (excludes deprecated types)\n VALID_STACK_TYPES = [\n StackType.BURST_SEQUENCE,\n StackType.EXPOSURE_BRACKET,\n StackType.MANUAL,\n ]\n\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n owner = models.ForeignKey(\n User,\n on_delete=models.SET(get_deleted_user),\n related_name=\"photo_stacks\",\n )\n\n stack_type = models.CharField(\n max_length=20,\n choices=StackType.choices,\n default=StackType.MANUAL,\n db_index=True,\n )\n\n # The photo shown in the timeline (cover photo)\n primary_photo = models.ForeignKey(\n \"Photo\",\n on_delete=models.SET_NULL,\n null=True,\n blank=True,\n related_name=\"primary_in_stack\",\n )\n\n # Detection metadata\n created_at = models.DateTimeField(auto_now_add=True)\n updated_at = models.DateTimeField(auto_now=True)\n\n # For bursts: time span of the sequence\n sequence_start = models.DateTimeField(null=True, blank=True)\n sequence_end = models.DateTimeField(null=True, blank=True)\n\n class Meta:\n ordering = [\"-created_at\"]\n verbose_name = \"Photo Stack\"\n verbose_name_plural = \"Photo Stacks\"\n indexes = [\n models.Index(fields=[\"owner\", \"stack_type\"]),\n ]\n\n def __str__(self):\n return f\"PhotoStack {self.id} - {self.stack_type} - {self.owner.username}\"\n\n @property\n def photo_count(self):\n \"\"\"Number of photos in this stack.\"\"\"\n return self.photos.count()\n\n def get_photos_ordered_by_quality(self):\n \"\"\"\n Returns photos in the stack ordered by quality metrics.\n Higher resolution and larger file size are considered better quality.\n \"\"\"\n return self.photos.order_by(\"-metadata__width\", \"-metadata__height\", \"-size\")\n\n def auto_select_primary(self):\n \"\"\"\n Automatically selects the best photo as primary based on stack type.\n\n For BURST_SEQUENCE: Middle of sequence by timestamp\n For EXPOSURE_BRACKET: Middle exposure\n For MANUAL: Highest resolution\n \n For deprecated types (RAW_JPEG_PAIR, LIVE_PHOTO):\n These should be migrated to Photo.files, but for compatibility:\n - RAW_JPEG_PAIR: Prefer JPEG (non-RAW)\n - LIVE_PHOTO: Prefer still image (non-video)\n \"\"\"\n photos = self.photos.all()\n if not photos.exists():\n return None\n\n if self.stack_type == self.StackType.BURST_SEQUENCE:\n # Pick middle of sequence by timestamp\n ordered = photos.order_by(\"exif_timestamp\")\n count = ordered.count()\n best = ordered[count // 2] if count > 0 else None\n elif self.stack_type == self.StackType.EXPOSURE_BRACKET:\n # Pick middle exposure (usually the \"correct\" exposure)\n ordered = photos.order_by(\"exif_timestamp\")\n count = ordered.count()\n best = ordered[count // 2] if count > 0 else None\n elif self.stack_type == self.StackType.RAW_JPEG_PAIR:\n # DEPRECATED: Prefer JPEG for display (RAW files have type=4)\n jpeg_photos = photos.exclude(main_file__type=4)\n best = jpeg_photos.first() if jpeg_photos.exists() else photos.first()\n elif self.stack_type == self.StackType.LIVE_PHOTO:\n # DEPRECATED: Prefer still image over video (VIDEO = 2)\n still_photos = photos.exclude(main_file__type=2)\n best = still_photos.first() if still_photos.exists() else photos.first()\n else:\n # MANUAL and default: highest resolution\n # Use metadata__width and metadata__height since these fields moved to PhotoMetadata\n best = photos.order_by(\n models.F(\"metadata__width\") * models.F(\"metadata__height\")\n ).last()\n\n if best:\n self.primary_photo = best\n self.save(update_fields=[\"primary_photo\", \"updated_at\"])\n\n return best\n\n def merge_with(self, other_stack: \"PhotoStack\"):\n \"\"\"\n Merge another stack into this one.\n All photos from the other stack are moved to this stack,\n and the other stack is deleted.\n \"\"\"\n if other_stack.pk == self.pk:\n return\n\n # Move all photos from other stack to this one (ManyToMany)\n # Convert to list first to avoid modifying queryset while iterating\n photos_to_move = list(other_stack.photos.all())\n for photo in photos_to_move:\n photo.stacks.add(self)\n photo.stacks.remove(other_stack)\n\n # Recalculate primary if needed\n if not self.primary_photo:\n self.auto_select_primary()\n\n # Delete the now-empty stack\n other_stack.delete()\n\n @classmethod\n def create_or_merge(cls, owner, stack_type, photos, sequence_start=None, sequence_end=None):\n \"\"\"\n Create a new stack or merge into existing if any photo is already stacked.\n\n Args:\n owner: User who owns the photos\n stack_type: Type of stack to create\n photos: Queryset or list of Photo objects to group\n sequence_start: Optional start timestamp for burst/bracket sequences\n sequence_end: Optional end timestamp for burst/bracket sequences\n\n Returns:\n The PhotoStack instance (new or existing)\n \"\"\"\n photo_list = list(photos)\n if len(photo_list) < 2:\n return None\n\n # Check if any photo is already in a stack of the same type\n existing_stacks = cls.objects.filter(\n photos__in=photo_list,\n stack_type=stack_type,\n owner=owner,\n ).distinct()\n\n if existing_stacks.exists():\n # Merge all into the first existing stack\n target_stack = existing_stacks.first()\n for stack in existing_stacks[1:]:\n target_stack.merge_with(stack)\n\n # Add any new photos to the stack (ManyToMany)\n for photo in photo_list:\n if not photo.stacks.filter(pk=target_stack.pk).exists():\n photo.stacks.add(target_stack)\n\n # Update sequence timestamps if provided and this is a burst/bracket stack\n if sequence_start is not None and sequence_end is not None:\n if target_stack.sequence_start is None or target_stack.sequence_start > sequence_start:\n target_stack.sequence_start = sequence_start\n if target_stack.sequence_end is None or target_stack.sequence_end < sequence_end:\n target_stack.sequence_end = sequence_end\n target_stack.save(update_fields=['sequence_start', 'sequence_end', 'updated_at'])\n\n target_stack.auto_select_primary()\n return target_stack\n else:\n # Create new stack\n stack = cls.objects.create(\n owner=owner,\n stack_type=stack_type,\n sequence_start=sequence_start,\n sequence_end=sequence_end,\n )\n\n # Associate photos (ManyToMany - add each photo to the stack)\n for photo in photo_list:\n photo.stacks.add(stack)\n\n stack.auto_select_primary()\n return stack\n" }, { "path": "api/models/stack_review.py", "content": "\"\"\"\nStackReview model for tracking user review decisions on reviewable stacks.\n\nThis model is separate from PhotoStack because:\n1. Only certain stack types need review (exact_copy, visual_duplicate)\n2. Other stack types (raw_jpeg, burst, bracket, live_photo) are informational\n3. Manual stacks are always \"reviewed\" by definition (user created them)\n\nThis separation allows:\n- Clean data model with clear semantics\n- Different workflows for different stack types\n- Historical tracking of review decisions\n\"\"\"\n\nimport uuid\n\nfrom django.db import models\n\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.user import User, get_deleted_user\n\n\nclass StackReview(models.Model):\n \"\"\"\n Records a user's review decision for a reviewable stack.\n \n Reviewable stack types:\n - exact_copy: User decides which copy to keep\n - visual_duplicate: User decides if photos are truly duplicates\n \n Non-reviewable stack types (informational only):\n - raw_jpeg: Both RAW and JPEG are kept (different purposes)\n - burst: User may want to browse all burst photos\n - bracket: HDR processing may need all exposures\n - live_photo: Photo and video are intrinsically linked\n - manual: User explicitly created the grouping\n \"\"\"\n\n class Decision(models.TextChoices):\n # User hasn't reviewed yet\n PENDING = \"pending\", \"Pending Review\"\n # User selected a primary and trashed others\n RESOLVED = \"resolved\", \"Resolved\"\n # User marked as \"not actually duplicates\"\n DISMISSED = \"dismissed\", \"Dismissed\"\n\n id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)\n\n stack = models.OneToOneField(\n PhotoStack,\n on_delete=models.CASCADE,\n related_name=\"review\",\n )\n\n reviewer = models.ForeignKey(\n User,\n on_delete=models.SET(get_deleted_user),\n related_name=\"stack_reviews\",\n )\n\n decision = models.CharField(\n max_length=20,\n choices=Decision.choices,\n default=Decision.PENDING,\n db_index=True,\n )\n\n # The photo the user chose to keep (only set when decision=RESOLVED)\n kept_photo = models.ForeignKey(\n \"Photo\",\n on_delete=models.SET_NULL,\n null=True,\n blank=True,\n related_name=\"kept_in_reviews\",\n )\n\n # Number of photos trashed when resolved\n trashed_count = models.IntegerField(default=0)\n\n # Timestamps\n created_at = models.DateTimeField(auto_now_add=True)\n reviewed_at = models.DateTimeField(null=True, blank=True)\n\n # Optional note from user\n note = models.TextField(blank=True, null=True)\n\n class Meta:\n ordering = [\"-created_at\"]\n verbose_name = \"Stack Review\"\n verbose_name_plural = \"Stack Reviews\"\n indexes = [\n models.Index(fields=[\"reviewer\", \"decision\"]),\n ]\n\n def __str__(self):\n return f\"Review for {self.stack.id} - {self.decision}\"\n\n @classmethod\n def is_reviewable_type(cls, stack_type: str) -> bool:\n \"\"\"\n Check if a stack type requires user review.\n \n Currently no stack types are reviewable because:\n - exact_copy and visual_duplicate are now handled by Duplicate model\n - BURST_SEQUENCE, EXPOSURE_BRACKET, MANUAL are informational\n - RAW_JPEG_PAIR and LIVE_PHOTO are deprecated (use Photo.files)\n \n Returns:\n False for all current stack types\n \"\"\"\n # No current stack types require review\n # Duplicates are handled by the separate Duplicate model\n return False\n\n @classmethod\n def create_for_stack(cls, stack: PhotoStack) -> \"StackReview | None\":\n \"\"\"\n Create a review record for a stack if it's a reviewable type.\n Returns None for non-reviewable stack types.\n \"\"\"\n if not cls.is_reviewable_type(stack.stack_type):\n return None\n\n review, created = cls.objects.get_or_create(\n stack=stack,\n defaults={\n \"reviewer\": stack.owner,\n \"decision\": cls.Decision.PENDING,\n }\n )\n return review\n\n def resolve(self, kept_photo, trash_others: bool = True):\n \"\"\"\n Mark the review as resolved by selecting a photo to keep.\n \n Args:\n kept_photo: The Photo instance to keep as primary\n trash_others: Whether to move other photos to trash\n \"\"\"\n from django.utils import timezone\n\n # Set the kept photo\n self.kept_photo = kept_photo\n self.decision = self.Decision.RESOLVED\n self.reviewed_at = timezone.now()\n\n # Also set as stack's primary photo\n self.stack.primary_photo = kept_photo\n self.stack.save(update_fields=[\"primary_photo\", \"updated_at\"])\n\n # Trash others if requested\n if trash_others:\n other_photos = self.stack.photos.exclude(pk=kept_photo.pk)\n self.trashed_count = other_photos.update(in_trashcan=True)\n \n self.save()\n return self\n\n def dismiss(self):\n \"\"\"Mark as 'not actually duplicates' and unlink photos from stack.\"\"\"\n from django.utils import timezone\n\n self.decision = self.Decision.DISMISSED\n self.reviewed_at = timezone.now()\n \n # Unlink photos from stack (ManyToMany)\n for photo in self.stack.photos.all():\n photo.stacks.remove(self.stack)\n \n self.save()\n return self\n\n def revert(self):\n \"\"\"Revert a resolved review, restoring trashed photos.\"\"\"\n if self.decision != self.Decision.RESOLVED:\n return self\n\n # Restore trashed photos in this stack (using ManyToMany relationship)\n restored_count = self.stack.photos.filter(\n in_trashcan=True\n ).update(in_trashcan=False)\n\n # Reset to pending\n self.decision = self.Decision.PENDING\n self.kept_photo = None\n self.trashed_count = 0\n self.reviewed_at = None\n \n # Clear stack's primary\n self.stack.primary_photo = None\n self.stack.save(update_fields=[\"primary_photo\", \"updated_at\"])\n\n self.save()\n return restored_count\n" }, { "path": "api/models/thumbnail.py", "content": "import os\n\nfrom django.db import models\nfrom PIL import Image\n\nfrom api.metadata.reader import get_metadata\nfrom api.metadata.tags import Tags\nfrom api.models.photo import Photo\nfrom api.thumbnails import (\n create_animated_thumbnail,\n create_thumbnail,\n create_thumbnail_for_video,\n does_static_thumbnail_exist,\n does_video_thumbnail_exist,\n)\nfrom api.util import logger\n\n\nclass Thumbnail(models.Model):\n photo = models.OneToOneField(\n Photo, on_delete=models.CASCADE, related_name=\"thumbnail\", primary_key=True\n )\n thumbnail_big = models.ImageField(upload_to=\"thumbnails_big\")\n square_thumbnail = models.ImageField(upload_to=\"square_thumbnails\")\n square_thumbnail_small = models.ImageField(upload_to=\"square_thumbnails_small\")\n aspect_ratio = models.FloatField(blank=True, null=True)\n dominant_color = models.TextField(blank=True, null=True)\n\n def _generate_thumbnail(self):\n try:\n # Use photo.image_hash for thumbnail paths for frontend compatibility\n photo_hash = self.photo.image_hash\n if not does_static_thumbnail_exist(\"thumbnails_big\", photo_hash):\n if not self.photo.video:\n create_thumbnail(\n input_path=self.photo.main_file.path,\n output_height=1080,\n output_path=\"thumbnails_big\",\n hash=photo_hash,\n file_type=\".webp\",\n )\n else:\n create_thumbnail_for_video(\n input_path=self.photo.main_file.path,\n output_path=\"thumbnails_big\",\n hash=photo_hash,\n file_type=\".webp\",\n )\n\n if not self.photo.video and not does_static_thumbnail_exist(\n \"square_thumbnails\", photo_hash\n ):\n create_thumbnail(\n input_path=self.photo.main_file.path,\n output_height=500,\n output_path=\"square_thumbnails\",\n hash=photo_hash,\n file_type=\".webp\",\n )\n if self.photo.video and not does_video_thumbnail_exist(\n \"square_thumbnails\", photo_hash\n ):\n create_animated_thumbnail(\n input_path=self.photo.main_file.path,\n output_height=500,\n output_path=\"square_thumbnails\",\n hash=photo_hash,\n file_type=\".mp4\",\n )\n\n if not self.photo.video and not does_static_thumbnail_exist(\n \"square_thumbnails_small\", photo_hash\n ):\n create_thumbnail(\n input_path=self.photo.main_file.path,\n output_height=250,\n output_path=\"square_thumbnails_small\",\n hash=photo_hash,\n file_type=\".webp\",\n )\n if self.photo.video and not does_video_thumbnail_exist(\n \"square_thumbnails_small\", photo_hash\n ):\n create_animated_thumbnail(\n input_path=self.photo.main_file.path,\n output_height=250,\n output_path=\"square_thumbnails_small\",\n hash=photo_hash,\n file_type=\".mp4\",\n )\n filetype = \".webp\"\n if self.photo.video:\n filetype = \".mp4\"\n self.thumbnail_big.name = os.path.join(\n \"thumbnails_big\", photo_hash + \".webp\"\n )\n self.square_thumbnail.name = os.path.join(\n \"square_thumbnails\", photo_hash + filetype\n )\n self.square_thumbnail_small.name = os.path.join(\n \"square_thumbnails_small\", photo_hash + filetype\n )\n self.save()\n except Exception as e:\n logger.exception(\n f\"could not generate thumbnail for image {self.photo.main_file.path}\"\n )\n raise e\n\n def _calculate_aspect_ratio(self):\n try:\n # Relies on big thumbnail for correct aspect ratio\n height, width = get_metadata(\n self.thumbnail_big.path,\n tags=[Tags.IMAGE_HEIGHT, Tags.IMAGE_WIDTH],\n try_sidecar=False,\n )\n self.aspect_ratio = round(width / height, 2)\n\n self.save()\n except Exception as e:\n logger.exception(\n f\"could not calculate aspect ratio for image {self.thumbnail_big.path}\"\n )\n raise e\n\n def _get_dominant_color(self, palette_size=16):\n # Skip if it's already calculated\n if self.dominant_color:\n return\n try:\n # Resize image to speed up processing\n img = Image.open(self.square_thumbnail_small.path)\n img.thumbnail((100, 100))\n\n # Reduce colors (uses k-means internally)\n paletted = img.convert(\"P\", palette=Image.ADAPTIVE, colors=palette_size)\n\n # Find the color that occurs most often\n palette = paletted.getpalette()\n color_counts = sorted(paletted.getcolors(), reverse=True)\n palette_index = color_counts[0][1]\n dominant_color = palette[palette_index * 3 : palette_index * 3 + 3]\n self.dominant_color = dominant_color\n self.save()\n except Exception:\n logger.info(f\"Cannot calculate dominant color {self} object\")\n" }, { "path": "api/models/user.py", "content": "import pytz\nfrom django.conf import settings\nfrom django.contrib.auth.models import AbstractUser\nfrom django.db import models\nfrom django_cryptography.fields import encrypt\n\nfrom api.date_time_extractor import DEFAULT_RULES_JSON\nfrom api.burst_detection_rules import get_default_burst_detection_rules\n\n\ndef get_default_config_datetime_rules(): # This is a callable\n return DEFAULT_RULES_JSON\n\n\ndef get_default_config_burst_detection_rules(): # This is a callable\n return get_default_burst_detection_rules()\n\n\ndef get_default_llm_settings():\n return {\n \"enabled\": False,\n \"add_person\": False,\n \"add_location\": False,\n \"add_keywords\": False,\n \"add_camera\": False,\n \"add_lens\": False,\n \"add_album\": False,\n \"sentiment\": 0,\n \"custom_prompt\": \"\",\n \"custom_prompt_enabled\": False,\n }\n\n\ndef get_default_public_sharing_settings():\n \"\"\"Default settings for what metadata to share in public albums.\n\n All options default to False (opt-in) for privacy-first approach.\n Users can change their defaults, and per-album overrides take precedence.\n \"\"\"\n return {\n \"share_location\": False, # GPS coordinates and location names\n \"share_camera_info\": False, # Camera make/model, lens, settings\n \"share_timestamps\": False, # Date/time when photo was taken\n \"share_captions\": False, # AI-generated or user captions\n \"share_faces\": False, # Detected faces (always recommend False)\n }\n\n\nclass User(AbstractUser):\n scan_directory = models.CharField(\n max_length=512, db_index=True, blank=True, default=\"\"\n )\n confidence = models.FloatField(default=0.1, db_index=True)\n confidence_person = models.FloatField(default=0.9)\n image_scale = models.FloatField(default=1)\n semantic_search_topk = models.IntegerField(default=0)\n avatar = models.ImageField(upload_to=\"avatars\", null=True, blank=True)\n transcode_videos = models.BooleanField(default=False)\n nextcloud_server_address = models.CharField(max_length=200, default=\"\", blank=True)\n nextcloud_username = models.CharField(max_length=64, default=\"\", blank=True)\n nextcloud_app_password = encrypt(\n models.CharField(max_length=64, default=\"\", blank=True)\n )\n nextcloud_scan_directory = models.CharField(\n max_length=512, db_index=True, default=\"\", blank=True\n )\n\n favorite_min_rating = models.IntegerField(\n default=settings.DEFAULT_FAVORITE_MIN_RATING, db_index=True\n )\n\n class SaveMetadata(models.TextChoices):\n OFF = \"OFF\"\n MEDIA_FILE = \"MEDIA_FILE\"\n SIDECAR_FILE = \"SIDECAR_FILE\"\n\n save_metadata_to_disk = models.TextField(\n choices=SaveMetadata.choices, default=SaveMetadata.OFF\n )\n save_face_tags_to_disk = models.BooleanField(default=False)\n llm_settings = models.JSONField(default=get_default_llm_settings)\n datetime_rules = models.JSONField(default=get_default_config_datetime_rules)\n burst_detection_rules = models.JSONField(\n default=get_default_config_burst_detection_rules\n )\n default_timezone = models.TextField(\n choices=[(x, x) for x in pytz.all_timezones],\n default=\"UTC\",\n )\n public_sharing = models.BooleanField(default=False)\n public_sharing_defaults = models.JSONField(\n default=get_default_public_sharing_settings\n )\n\n class FaceRecogniton(models.TextChoices):\n HOG = \"HOG\"\n CNN = \"CNN\"\n\n face_recognition_model = models.TextField(\n choices=FaceRecogniton.choices, default=FaceRecogniton.HOG\n )\n min_cluster_size = models.IntegerField(default=0)\n confidence_unknown_face = models.FloatField(default=0.5)\n min_samples = models.IntegerField(default=1)\n cluster_selection_epsilon = models.FloatField(default=0.05)\n\n class TextAlignment(models.TextChoices):\n LEFT = \"left\"\n RIGHT = \"right\"\n\n text_alignment = models.TextField(\n choices=TextAlignment.choices, default=TextAlignment.RIGHT\n )\n\n class HeaderSize(models.TextChoices):\n LARGE = \"large\"\n NORMAL = \"normal\"\n SMALL = \"small\"\n\n header_size = models.TextField(choices=HeaderSize.choices, default=HeaderSize.LARGE)\n skip_raw_files = models.BooleanField(\n default=False\n ) # Deprecated: kept for migration compatibility\n stack_raw_jpeg = models.BooleanField(default=True)\n slideshow_interval = models.IntegerField(default=5)\n\n # Duplicate detection settings\n class DuplicateSensitivity(models.TextChoices):\n STRICT = \"strict\"\n NORMAL = \"normal\"\n LOOSE = \"loose\"\n\n duplicate_sensitivity = models.TextField(\n choices=DuplicateSensitivity.choices, default=DuplicateSensitivity.NORMAL\n )\n duplicate_clear_existing = models.BooleanField(default=False)\n\n\ndef get_admin_user():\n return User.objects.get(is_superuser=True)\n\n\ndef get_deleted_user():\n deleted_user: User = User.objects.get_or_create(username=\"deleted\")[0]\n if deleted_user.is_active is not False:\n deleted_user.is_active = False\n deleted_user.save()\n return deleted_user\n" }, { "path": "api/nextcloud.py", "content": "import os\n\nimport owncloud as nextcloud\n\n\ndef login(user):\n nc = nextcloud.Client(user.nextcloud_server_address)\n nc.login(user.nextcloud_username, user.nextcloud_app_password)\n\n def path_to_dict(path):\n d = {\"title\": os.path.basename(path), \"absolute_path\": path}\n try:\n d[\"children\"] = [\n path_to_dict(os.path.join(path, x.path))\n for x in nc.list(path)\n if x.is_dir()\n ]\n except Exception:\n pass\n\n return d\n\n\ndef list_dir(user, path):\n nc = nextcloud.Client(user.nextcloud_server_address)\n nc.login(user.nextcloud_username, user.nextcloud_app_password)\n return [p.path for p in nc.list(path) if p.is_dir()]\n" }, { "path": "api/perceptual_hash.py", "content": "\"\"\"\nPerceptual hashing module for duplicate image detection.\n\nUses pHash (perceptual hash) algorithm which is robust to:\n- Image resizing/scaling\n- Compression artifacts (JPEG quality differences)\n- Minor color adjustments\n- Small crops (up to ~15% border removal)\n\"\"\"\n\nimport imagehash\nfrom PIL import Image\n\nfrom api.util import logger\n\n# Threshold for considering two images as duplicates\n# pHash produces 64-bit hashes, Hamming distance <= 10 indicates high similarity\nDEFAULT_HAMMING_THRESHOLD = 10\n\n\ndef calculate_perceptual_hash(image_path: str, hash_size: int = 8) -> str | None:\n \"\"\"\n Calculate the perceptual hash (pHash) of an image.\n\n Args:\n image_path: Path to the image file\n hash_size: Size of the hash (default 8 produces 64-bit hash)\n\n Returns:\n Hex string representation of the perceptual hash, or None if failed\n \"\"\"\n try:\n with Image.open(image_path) as img:\n # Convert to RGB if necessary (handles RGBA, palette images, etc.)\n if img.mode not in (\"RGB\", \"L\"):\n img = img.convert(\"RGB\")\n\n # Calculate pHash - uses DCT (Discrete Cosine Transform)\n # More robust than average hash or difference hash\n phash = imagehash.phash(img, hash_size=hash_size)\n return str(phash)\n except Exception as e:\n logger.error(f\"Failed to calculate perceptual hash for {image_path}: {e}\")\n return None\n\n\ndef calculate_hash_from_thumbnail(thumbnail_path: str) -> str | None:\n \"\"\"\n Calculate perceptual hash from a thumbnail image.\n Using thumbnails is faster and still produces reliable hashes.\n\n Args:\n thumbnail_path: Path to the thumbnail file\n\n Returns:\n Hex string representation of the perceptual hash, or None if failed\n \"\"\"\n return calculate_perceptual_hash(thumbnail_path)\n\n\ndef hamming_distance(hash1: str, hash2: str) -> int:\n \"\"\"\n Calculate the Hamming distance between two perceptual hashes.\n\n Args:\n hash1: First hash as hex string\n hash2: Second hash as hex string\n\n Returns:\n Number of differing bits (0 = identical, higher = more different)\n \"\"\"\n try:\n h1 = imagehash.hex_to_hash(hash1)\n h2 = imagehash.hex_to_hash(hash2)\n return h1 - h2 # imagehash overloads subtraction to return Hamming distance\n except Exception as e:\n logger.error(f\"Failed to calculate Hamming distance: {e}\")\n return 64 # Maximum distance (completely different)\n\n\ndef are_duplicates(hash1: str, hash2: str, threshold: int = DEFAULT_HAMMING_THRESHOLD) -> bool:\n \"\"\"\n Determine if two images are duplicates based on their perceptual hashes.\n\n Args:\n hash1: First perceptual hash\n hash2: Second perceptual hash\n threshold: Maximum Hamming distance to consider as duplicates (default 10)\n\n Returns:\n True if images are likely duplicates, False otherwise\n \"\"\"\n if not hash1 or not hash2:\n return False\n return hamming_distance(hash1, hash2) <= threshold\n\n\ndef find_similar_hashes(\n target_hash: str,\n hash_list: list[tuple[str, str]],\n threshold: int = DEFAULT_HAMMING_THRESHOLD,\n) -> list[tuple[str, int]]:\n \"\"\"\n Find all hashes similar to the target hash.\n\n Args:\n target_hash: The hash to compare against\n hash_list: List of (image_id, hash) tuples to search\n threshold: Maximum Hamming distance for similarity\n\n Returns:\n List of (image_id, distance) tuples for similar images, sorted by distance\n \"\"\"\n if not target_hash:\n return []\n\n similar = []\n for image_id, hash_value in hash_list:\n if hash_value and hash_value != target_hash:\n distance = hamming_distance(target_hash, hash_value)\n if distance <= threshold:\n similar.append((image_id, distance))\n\n return sorted(similar, key=lambda x: x[1])\n" }, { "path": "api/permissions.py", "content": "from constance import config as site_config\nfrom rest_framework import permissions\n\nfrom api.models import User\n\n\nclass IsAdminOrSelf(permissions.BasePermission):\n def has_object_permission(self, request, view, obj):\n if request.method in permissions.SAFE_METHODS:\n return True\n\n return request.user and request.user.is_staff or obj == request.user\n\n\nclass IsAdminOrFirstTimeSetupOrRegistrationAllowed(permissions.BasePermission):\n def has_permission(self, request, view):\n if request.method in permissions.SAFE_METHODS:\n return True\n\n is_admin = request.user and request.user.is_staff\n is_first_time_setup = not User.objects.filter(is_superuser=True).exists()\n is_registration_allowed = bool(site_config.ALLOW_REGISTRATION)\n\n return is_admin or is_first_time_setup or is_registration_allowed\n\n\nclass IsOwnerOrReadOnly(permissions.BasePermission):\n \"\"\"Custom permission to only allow owners of an object to edit it.\"\"\"\n\n def has_object_permission(self, request, view, obj):\n # Read permissions are allowed to any request,\n # so we'll always allow GET, HEAD or OPTIONS requests.\n if request.method in permissions.SAFE_METHODS:\n return True\n\n # Write permissions are only allowed to the owner of the snippet.\n return obj.owner == request.user\n\n\nclass IsUserOrReadOnly(permissions.BasePermission):\n \"\"\"Custom permission to only allow owners of an object to edit it.\"\"\"\n\n def has_object_permission(self, request, view, obj):\n # Read permissions are allowed to any request,\n # so we'll always allow GET, HEAD or OPTIONS requests.\n if request.method in permissions.SAFE_METHODS:\n return True\n\n # Write permissions are only allowed to the owner of the snippet.\n return obj == request.user\n\n\nclass IsPhotoOrAlbumSharedTo(permissions.BasePermission):\n \"\"\"Custom permission to only allow owners of an object to edit it.\"\"\"\n\n def has_object_permission(self, request, view, obj):\n if obj.public:\n return True\n\n if obj.owner == request.user or request.user in obj.shared_to.all():\n return True\n\n for album in obj.albumuser_set.only(\"shared_to\"):\n if request.user in album.shared_to.all():\n return True\n\n return False\n\n\nclass IsRegistrationAllowed(permissions.BasePermission):\n \"\"\"Custom permission to only allow if registration is allowed globally.\"\"\"\n\n def has_permission(self, request, view):\n return bool(site_config.ALLOW_REGISTRATION)\n" }, { "path": "api/schemas/site_settings.py", "content": "site_settings_schema = {\n \"type\": \"object\",\n \"anyOf\": [\n {\"required\": [\"allow_registration\"]},\n {\"required\": [\"allow_upload\"]},\n {\"required\": [\"skip_patterns\"]},\n {\"required\": [\"map_api_provider\"]},\n {\"required\": [\"map_api_key\"]},\n {\"required\": [\"captioning_model\"]},\n {\"required\": [\"llm_model\"]},\n {\"required\": [\"tagging_model\"]},\n ],\n \"properties\": {\n \"allow_registration\": {\"type\": \"boolean\"},\n \"allow_upload\": {\"type\": \"boolean\"},\n \"skip_patterns\": {\"type\": \"string\"},\n \"map_api_provider\": {\"type\": \"string\"},\n \"map_api_key\": {\"type\": \"string\"},\n \"captioning_model\": {\"type\": \"string\"},\n \"llm_model\": {\"type\": \"string\"},\n \"tagging_model\": {\"type\": \"string\"},\n },\n}\n" }, { "path": "api/semantic_search.py", "content": "import numpy as np\nimport requests\nfrom django.conf import settings\n\ndir_clip_ViT_B_32_model = settings.CLIP_ROOT\n\n\ndef create_clip_embeddings(imgs):\n json = {\n \"imgs\": imgs,\n \"model\": dir_clip_ViT_B_32_model,\n }\n clip_embeddings = requests.post(\n \"http://localhost:8006/clip-embeddings\", json=json\n ).json()\n\n imgs_emb = clip_embeddings[\"imgs_emb\"]\n magnitudes = clip_embeddings[\"magnitudes\"]\n\n # Convert Python lists to NumPy arrays\n imgs_emb = [np.array(enc) for enc in imgs_emb]\n\n return imgs_emb, magnitudes\n\n\ndef calculate_query_embeddings(query):\n json = {\n \"query\": query,\n \"model\": dir_clip_ViT_B_32_model,\n }\n query_embedding = requests.post(\n \"http://localhost:8006/query-embeddings\", json=json\n ).json()\n\n emb = query_embedding[\"emb\"]\n magnitude = query_embedding[\"magnitude\"]\n return emb, magnitude\n" }, { "path": "api/serializers/PhotosGroupedByDate.py", "content": "import pytz\nfrom itertools import groupby\n\nutc = pytz.UTC\n\n\nclass PhotosGroupedByDate:\n def __init__(self, location, date, photos):\n self.photos = photos\n self.date = date\n self.location = location\n\n\ndef get_photos_ordered_by_date(photos):\n \"\"\"\n Efficiently group photos by date using itertools.groupby.\n Assumes photos are already ordered by exif_timestamp.\n \"\"\"\n # Convert to list once if it's a queryset\n if hasattr(photos, \"_result_cache\") and photos._result_cache is None:\n photos = list(photos)\n\n result = []\n no_timestamp_photos = []\n\n def date_key(photo):\n \"\"\"Key function for grouping photos by date\"\"\"\n if photo.exif_timestamp:\n return photo.exif_timestamp.date().strftime(\"%Y-%m-%d\")\n return None\n\n # Group consecutive photos by their date\n for date_str, group_photos in groupby(photos, key=date_key):\n group_list = list(group_photos)\n location = \"\"\n\n if date_str is not None:\n # Use the first photo's timestamp as the group date\n date = group_list[0].exif_timestamp\n result.append(PhotosGroupedByDate(location, date, group_list))\n else:\n # Collect photos without timestamps\n no_timestamp_photos.extend(group_list)\n\n # Add no timestamp photos as a single group at the end\n if no_timestamp_photos:\n result.append(PhotosGroupedByDate(\"\", \"No timestamp\", no_timestamp_photos))\n\n return result\n" }, { "path": "api/serializers/__init__.py", "content": "" }, { "path": "api/serializers/album_auto.py", "content": "from rest_framework import serializers\n\nfrom api.models import AlbumAuto\nfrom api.serializers.person import PersonSerializer\nfrom api.serializers.photos import PhotoHashListSerializer\nfrom api.serializers.simple import PhotoSimpleSerializer\n\n\nclass AlbumAutoSerializer(serializers.ModelSerializer):\n photos = PhotoSimpleSerializer(many=True, read_only=False)\n people = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumAuto\n fields = (\n \"id\",\n \"title\",\n \"favorited\",\n \"timestamp\",\n \"created_on\",\n \"gps_lat\",\n \"people\",\n \"gps_lon\",\n \"photos\",\n )\n\n def get_people(self, obj) -> PersonSerializer(many=True):\n res = []\n for photo in obj.photos.all():\n faces = photo.faces.all()\n for face in faces:\n serialized_person = PersonSerializer(face.person).data\n if serialized_person not in res:\n res.append(serialized_person)\n return res\n\n def delete(self, validated_data, id):\n album = AlbumAuto.objects.filter(id=id).get()\n album.delete()\n\n\nclass AlbumAutoListSerializer(serializers.ModelSerializer):\n photos = serializers.SerializerMethodField()\n photo_count = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumAuto\n fields = (\n \"id\",\n \"title\",\n \"timestamp\",\n \"photos\",\n \"photo_count\",\n \"favorited\",\n )\n\n def get_photo_count(self, obj) -> int:\n try:\n return obj.photo_count\n except Exception:\n return obj.photos.count()\n\n def get_photos(self, obj) -> PhotoHashListSerializer:\n try:\n return PhotoHashListSerializer(obj.cover_photo[0]).data\n except Exception:\n return \"\"\n" }, { "path": "api/serializers/album_date.py", "content": "from rest_framework import serializers\n\nfrom api.models import AlbumDate\n\n\nclass IncompleteAlbumDateSerializer(serializers.ModelSerializer):\n id = serializers.SerializerMethodField()\n date = serializers.SerializerMethodField()\n location = serializers.SerializerMethodField()\n incomplete = serializers.SerializerMethodField()\n numberOfItems = serializers.SerializerMethodField(\"get_number_of_items\")\n items = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumDate\n fields = (\"id\", \"date\", \"location\", \"incomplete\", \"numberOfItems\", \"items\")\n\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def get_date(self, obj) -> str:\n if obj.date:\n return obj.date.isoformat()\n else:\n return None\n\n def get_items(self, obj) -> list:\n return []\n\n def get_incomplete(self, obj) -> bool:\n return True\n\n def get_number_of_items(self, obj) -> int:\n if obj and obj.photo_count:\n return obj.photo_count\n else:\n return 0\n\n def get_location(self, obj) -> str:\n if obj and obj.location:\n return obj.location[\"places\"][0]\n else:\n return \"\"\n\n\nclass AlbumDateSerializer(serializers.ModelSerializer):\n id = serializers.SerializerMethodField()\n date = serializers.SerializerMethodField()\n location = serializers.SerializerMethodField()\n incomplete = serializers.SerializerMethodField()\n numberOfItems = serializers.SerializerMethodField(\"get_number_of_items\")\n items = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumDate\n fields = (\"id\", \"date\", \"location\", \"incomplete\", \"numberOfItems\", \"items\")\n\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def get_date(self, obj) -> str:\n if obj.date:\n return obj.date.isoformat()\n else:\n return None\n\n def get_items(self, obj) -> dict:\n # This method is removed as we're directly including paginated photos in the response.\n pass\n\n def get_incomplete(self, obj) -> bool:\n return False\n\n def get_number_of_items(self, obj) -> int:\n # this will also get added in the response\n pass\n\n def get_location(self, obj) -> str:\n if obj and obj.location:\n return obj.location[\"places\"][0]\n else:\n return \"\"\n" }, { "path": "api/serializers/album_place.py", "content": "from rest_framework import serializers\n\nfrom api.models import AlbumPlace\nfrom api.serializers.photos import GroupedPhotosSerializer, PhotoHashListSerializer\nfrom api.serializers.PhotosGroupedByDate import get_photos_ordered_by_date\nfrom api.serializers.simple import PhotoSuperSimpleSerializer\n\n\nclass GroupedPlacePhotosSerializer(serializers.ModelSerializer):\n id = serializers.SerializerMethodField()\n grouped_photos = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumPlace\n fields = (\n \"id\",\n \"title\",\n \"grouped_photos\",\n )\n\n # To-Do: Remove legacy stuff\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def get_grouped_photos(self, obj) -> GroupedPhotosSerializer(many=True):\n grouped_photos = get_photos_ordered_by_date(obj.photos.all())\n res = GroupedPhotosSerializer(grouped_photos, many=True).data\n return res\n\n\nclass AlbumPlaceSerializer(serializers.ModelSerializer):\n photos = PhotoSuperSimpleSerializer(many=True, read_only=True)\n\n class Meta:\n model = AlbumPlace\n fields = (\"id\", \"title\", \"photos\")\n\n\nclass AlbumPlaceListSerializer(serializers.ModelSerializer):\n cover_photos = PhotoHashListSerializer(many=True, read_only=True)\n photo_count = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumPlace\n fields = (\"id\", \"geolocation_level\", \"cover_photos\", \"title\", \"photo_count\")\n\n def get_photo_count(self, obj) -> int:\n return obj.photo_count\n" }, { "path": "api/serializers/album_thing.py", "content": "from rest_framework import serializers\n\nfrom api.models import AlbumThing\nfrom api.serializers.photos import GroupedPhotosSerializer, PhotoHashListSerializer\nfrom api.serializers.PhotosGroupedByDate import get_photos_ordered_by_date\nfrom api.serializers.simple import PhotoSuperSimpleSerializer\n\n\nclass GroupedThingPhotosSerializer(serializers.ModelSerializer):\n id = serializers.SerializerMethodField()\n grouped_photos = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumThing\n fields = (\n \"id\",\n \"title\",\n \"grouped_photos\",\n )\n\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def get_grouped_photos(self, obj) -> GroupedPhotosSerializer(many=True):\n grouped_photos = get_photos_ordered_by_date(obj.photos.all())\n res = GroupedPhotosSerializer(grouped_photos, many=True).data\n return res\n\n\nclass AlbumThingSerializer(serializers.ModelSerializer):\n photos = PhotoSuperSimpleSerializer(many=True, read_only=True)\n\n class Meta:\n model = AlbumThing\n fields = (\"id\", \"title\", \"photos\")\n\n\nclass AlbumThingListSerializer(serializers.ModelSerializer):\n cover_photos = PhotoHashListSerializer(many=True, read_only=True)\n photo_count = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumThing\n fields = (\n \"id\",\n \"cover_photos\",\n \"title\",\n \"photo_count\",\n \"thing_type\",\n \"cover_photos\",\n )\n\n def get_photo_count(self, obj) -> int:\n return obj.photo_count\n" }, { "path": "api/serializers/album_user.py", "content": "from rest_framework import serializers\n\nfrom api.models import AlbumUser, Photo\nfrom api.serializers.photos import GroupedPhotosSerializer\nfrom api.serializers.PhotosGroupedByDate import get_photos_ordered_by_date\nfrom api.serializers.simple import PhotoSuperSimpleSerializer, SimpleUserSerializer\nfrom api.util import logger\n\n\nclass AlbumUserSerializer(serializers.ModelSerializer):\n id = serializers.SerializerMethodField()\n owner = SimpleUserSerializer(many=False, read_only=True)\n shared_to = SimpleUserSerializer(many=True, read_only=True)\n date = serializers.SerializerMethodField()\n location = serializers.SerializerMethodField()\n grouped_photos = serializers.SerializerMethodField()\n public = serializers.SerializerMethodField()\n public_slug = serializers.SerializerMethodField()\n public_expires_at = serializers.SerializerMethodField()\n public_sharing_options = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumUser\n fields = (\n \"id\",\n \"title\",\n \"owner\",\n \"shared_to\",\n \"date\",\n \"location\",\n \"grouped_photos\",\n \"public\",\n \"public_slug\",\n \"public_expires_at\",\n \"public_sharing_options\",\n )\n\n # To-Do: Legacy definition, should be a number instead\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def get_grouped_photos(self, obj) -> GroupedPhotosSerializer(many=True):\n grouped_photos = get_photos_ordered_by_date(\n obj.photos.all().order_by(\"-exif_timestamp\")\n )\n res = GroupedPhotosSerializer(grouped_photos, many=True).data\n return res\n\n def get_location(self, obj) -> str:\n for photo in obj.photos.all():\n if (\n photo\n and hasattr(photo, \"search_instance\")\n and photo.search_instance\n and photo.search_instance.search_location\n ):\n return photo.search_instance.search_location\n return \"\"\n\n def get_date(self, obj) -> str:\n for photo in obj.photos.all():\n if photo and photo.exif_timestamp:\n return photo.exif_timestamp\n return \"\"\n\n def get_public(self, obj) -> bool:\n return bool(getattr(obj, \"share\", None) and obj.share.enabled)\n\n def get_public_slug(self, obj) -> str:\n return getattr(getattr(obj, \"share\", None), \"slug\", \"\") or \"\"\n\n def get_public_expires_at(self, obj):\n return getattr(getattr(obj, \"share\", None), \"expires_at\", None)\n\n def get_public_sharing_options(self, obj) -> dict | None:\n \"\"\"Return the per-album sharing option overrides (None values = use default).\"\"\"\n share = getattr(obj, \"share\", None)\n if not share:\n return None\n return {\n \"share_location\": share.share_location,\n \"share_camera_info\": share.share_camera_info,\n \"share_timestamps\": share.share_timestamps,\n \"share_captions\": share.share_captions,\n \"share_faces\": share.share_faces,\n }\n\n\nclass AlbumUserEditSerializer(serializers.ModelSerializer):\n photos = serializers.PrimaryKeyRelatedField(\n many=True, read_only=False, queryset=Photo.objects.all()\n )\n removedPhotos = serializers.ListField(\n child=serializers.CharField(max_length=100, default=\"\"),\n write_only=True,\n required=False,\n )\n\n class Meta:\n model = AlbumUser\n fields = (\n \"id\",\n \"title\",\n \"photos\",\n \"created_on\",\n \"favorited\",\n \"removedPhotos\",\n \"cover_photo\",\n )\n\n def create(self, validated_data):\n title = validated_data[\"title\"]\n photos = validated_data[\"photos\"]\n\n user = None\n request = self.context.get(\"request\")\n if request and hasattr(request, \"user\"):\n user = request.user\n\n # check if an album exists with the given title and call the update method if it does\n instance, created = AlbumUser.objects.get_or_create(title=title, owner=user)\n if not created:\n return self.update(instance, validated_data)\n\n for photo in photos:\n instance.photos.add(photo)\n instance.save()\n logger.info(f\"Created user album {instance.id} with {len(photos)} photos\")\n return instance\n\n def update(self, instance, validated_data):\n if \"title\" in validated_data.keys():\n title = validated_data[\"title\"]\n instance.title = title\n logger.info(f\"Renamed user album to {title}\")\n\n if \"removedPhotos\" in validated_data.keys():\n image_hashes = validated_data[\"removedPhotos\"]\n photos_already_in_album = instance.photos.all()\n cnt = 0\n for obj in photos_already_in_album:\n if obj.image_hash in image_hashes:\n cnt += 1\n instance.photos.remove(obj)\n\n logger.info(f\"Removed {cnt} photos to user album {instance.id}\")\n\n if \"cover_photo\" in validated_data.keys():\n cover_photo = validated_data[\"cover_photo\"]\n instance.cover_photo = cover_photo\n logger.info(f\"Changed cover photo to {cover_photo}\")\n\n if \"photos\" in validated_data.keys():\n photos = validated_data[\"photos\"]\n photos_already_in_album = instance.photos.all()\n cnt = 0\n for photo in photos:\n if photo not in photos_already_in_album:\n cnt += 1\n instance.photos.add(photo)\n\n logger.info(f\"Added {cnt} photos to user album {instance.id}\")\n\n instance.save()\n return instance\n\n\nclass AlbumUserListSerializer(serializers.ModelSerializer):\n cover_photo = serializers.SerializerMethodField()\n photo_count = serializers.SerializerMethodField()\n shared_to = SimpleUserSerializer(many=True, read_only=True)\n owner = SimpleUserSerializer(many=False, read_only=True)\n public = serializers.SerializerMethodField()\n public_slug = serializers.SerializerMethodField()\n public_expires_at = serializers.SerializerMethodField()\n public_sharing_options = serializers.SerializerMethodField()\n\n class Meta:\n model = AlbumUser\n fields = (\n \"id\",\n \"cover_photo\",\n \"created_on\",\n \"favorited\",\n \"title\",\n \"shared_to\",\n \"owner\",\n \"photo_count\",\n \"public\",\n \"public_slug\",\n \"public_expires_at\",\n \"public_sharing_options\",\n )\n\n def get_cover_photo(self, obj) -> PhotoSuperSimpleSerializer:\n if obj.cover_photo:\n return PhotoSuperSimpleSerializer(obj.cover_photo).data\n return PhotoSuperSimpleSerializer(obj.photos.first()).data\n\n def get_photo_count(self, obj) -> int:\n try:\n return obj.photo_count\n except Exception: # for when calling AlbumUserListSerializer(obj).data directly\n return obj.photos.count()\n\n def get_public(self, obj) -> bool:\n return bool(getattr(obj, \"share\", None) and obj.share.enabled)\n\n def get_public_slug(self, obj) -> str:\n return getattr(getattr(obj, \"share\", None), \"slug\", \"\") or \"\"\n\n def get_public_expires_at(self, obj):\n return getattr(getattr(obj, \"share\", None), \"expires_at\", None)\n\n def get_public_sharing_options(self, obj) -> dict | None:\n \"\"\"Return the per-album sharing option overrides (None values = use default).\"\"\"\n share = getattr(obj, \"share\", None)\n if not share:\n return None\n return {\n \"share_location\": share.share_location,\n \"share_camera_info\": share.share_camera_info,\n \"share_timestamps\": share.share_timestamps,\n \"share_captions\": share.share_captions,\n \"share_faces\": share.share_faces,\n }\n\n\nclass AlbumUserPublicSerializer(serializers.ModelSerializer):\n \"\"\"Serializer for publicly shared user albums.\n\n Ensures photos are grouped from a filtered subset (not hidden, not in trashcan).\n \"\"\"\n\n id = serializers.SerializerMethodField()\n owner = SimpleUserSerializer(many=False, read_only=True)\n date = serializers.SerializerMethodField()\n location = serializers.SerializerMethodField()\n grouped_photos = serializers.SerializerMethodField()\n public_slug = serializers.CharField(read_only=True)\n public_expires_at = serializers.DateTimeField(read_only=True)\n\n class Meta:\n model = AlbumUser\n fields = (\n \"id\",\n \"title\",\n \"owner\",\n \"date\",\n \"location\",\n \"grouped_photos\",\n \"public_slug\",\n \"public_expires_at\",\n )\n\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def _filtered_photos(self, obj):\n # Public albums should not expose hidden or trash photos\n return obj.photos.filter(hidden=False, in_trashcan=False).order_by(\n \"-exif_timestamp\"\n )\n\n def get_grouped_photos(self, obj) -> GroupedPhotosSerializer(many=True):\n grouped_photos = get_photos_ordered_by_date(self._filtered_photos(obj))\n return GroupedPhotosSerializer(grouped_photos, many=True).data\n\n def get_location(self, obj) -> str:\n for photo in self._filtered_photos(obj):\n if (\n photo\n and hasattr(photo, \"search_instance\")\n and photo.search_instance\n and photo.search_instance.search_location\n ):\n return photo.search_instance.search_location\n return \"\"\n\n def get_date(self, obj) -> str:\n for photo in self._filtered_photos(obj):\n if photo and photo.exif_timestamp:\n return photo.exif_timestamp\n return \"\"\n" }, { "path": "api/serializers/face.py", "content": "from rest_framework import serializers\n\nfrom api.models import Face, Person\n\n\nclass PersonFaceListSerializer(serializers.ModelSerializer):\n face_url = serializers.SerializerMethodField()\n person_label_probability = serializers.SerializerMethodField()\n photo_image_hash = serializers.SerializerMethodField()\n\n class Meta:\n model = Face\n fields = [\n \"id\",\n \"image\",\n \"face_url\",\n \"photo\",\n \"photo_image_hash\",\n \"timestamp\",\n \"person_label_probability\",\n ]\n\n def get_person_label_probability(self, obj):\n if obj.analysis_method == \"clustering\":\n return obj.cluster_probability\n else:\n return obj.classification_probability\n\n def get_face_url(self, obj):\n return obj.image.url\n\n def get_photo_image_hash(self, obj):\n return obj.photo.image_hash if obj.photo else None\n\n\nclass IncompletePersonFaceListSerializer(serializers.ModelSerializer):\n face_count = serializers.SerializerMethodField()\n\n class Meta:\n model = Person\n fields = [\"id\", \"name\", \"kind\", \"face_count\"]\n\n def get_face_count(self, obj) -> int:\n if obj and obj.viewable_face_count:\n return obj.viewable_face_count\n else:\n return 0\n\n\nclass FaceListSerializer(serializers.ModelSerializer):\n person_name = serializers.SerializerMethodField()\n face_url = serializers.SerializerMethodField()\n person_label_probability = serializers.SerializerMethodField()\n\n class Meta:\n model = Face\n fields = (\n \"id\",\n \"image\",\n \"face_url\",\n \"photo\",\n \"timestamp\",\n \"person\",\n \"person_label_probability\",\n \"person_name\",\n )\n\n def get_person_label_probability(self, obj) -> float:\n return obj.cluster_probability\n\n def get_face_url(self, obj) -> str:\n return obj.image.url\n\n def get_person_name(self, obj) -> str:\n if obj.person:\n return obj.person.name\n else:\n return \"Unknown - Other\"\n" }, { "path": "api/serializers/job.py", "content": "from rest_framework import serializers\n\nfrom api.models import LongRunningJob\nfrom api.serializers.simple import SimpleUserSerializer\n\n\nclass LongRunningJobSerializer(serializers.ModelSerializer):\n job_type_str = serializers.SerializerMethodField()\n started_by = SimpleUserSerializer(read_only=True)\n\n class Meta:\n model = LongRunningJob\n fields = (\n \"job_id\",\n \"queued_at\",\n \"finished\",\n \"finished_at\",\n \"started_at\",\n \"failed\",\n \"job_type_str\",\n \"job_type\",\n \"started_by\",\n \"progress_current\",\n \"progress_target\",\n \"progress_step\",\n \"result\",\n \"id\",\n )\n\n def get_job_type_str(self, obj) -> str:\n return dict(LongRunningJob.JOB_TYPES)[obj.job_type]\n" }, { "path": "api/serializers/person.py", "content": "from django.db.models import Q\nfrom rest_framework import serializers\n\nfrom api.models import Person, Photo\nfrom api.serializers.photos import GroupedPhotosSerializer\nfrom api.serializers.PhotosGroupedByDate import get_photos_ordered_by_date\nfrom api.util import logger\n\n\nclass GroupedPersonPhotosSerializer(serializers.ModelSerializer):\n id = serializers.SerializerMethodField()\n grouped_photos = serializers.SerializerMethodField()\n\n class Meta:\n model = Person\n fields = (\n \"id\",\n \"name\",\n \"grouped_photos\",\n )\n\n def get_id(self, obj) -> str:\n return str(obj.id)\n\n def get_grouped_photos(self, obj) -> GroupedPhotosSerializer(many=True):\n user = None\n request = self.context.get(\"request\")\n if request and hasattr(request, \"user\"):\n user = request.user\n grouped_photos = get_photos_ordered_by_date(obj.get_photos(user))\n res = GroupedPhotosSerializer(grouped_photos, many=True).data\n return res\n\n\nclass PersonSerializer(serializers.ModelSerializer):\n face_url = serializers.SerializerMethodField()\n face_photo_url = serializers.SerializerMethodField()\n video = serializers.SerializerMethodField()\n newPersonName = serializers.CharField(max_length=100, default=\"\", write_only=True)\n cover_photo = serializers.CharField(max_length=100, default=\"\", write_only=True)\n\n class Meta:\n model = Person\n fields = (\n \"name\",\n \"face_url\",\n \"face_count\",\n \"face_photo_url\",\n \"video\",\n \"id\",\n \"newPersonName\",\n \"cover_photo\",\n )\n\n def get_face_url(self, obj) -> str:\n if obj.cover_face:\n return \"/media/\" + obj.cover_face.image.name\n if obj.faces.count() == 0:\n return \"\"\n return \"/media/\" + obj.faces.first().image.name\n\n def get_face_photo_url(self, obj) -> str:\n if obj.cover_photo:\n return obj.cover_photo.image_hash\n if obj.faces.count() == 0:\n return \"\"\n return obj.faces.first().photo.image_hash\n\n def get_video(self, obj) -> str:\n if obj.cover_photo:\n return obj.cover_photo.video\n if obj.faces.count() == 0:\n return \"False\"\n return obj.faces.first().photo.video\n\n def create(self, validated_data):\n name = validated_data.pop(\"name\")\n if len(name.strip()) == 0:\n raise serializers.ValidationError(\"Name cannot be empty\")\n qs = Person.objects.filter(name=name)\n if qs.count() > 0:\n return qs[0]\n else:\n new_person = Person()\n new_person.name = name\n new_person.save()\n logger.info(f\"created person {new_person.id}\")\n return new_person\n\n def update(self, instance, validated_data):\n if \"newPersonName\" in validated_data.keys():\n new_name = validated_data.pop(\"newPersonName\")\n instance.name = new_name\n instance.save()\n return instance\n if \"cover_photo\" in validated_data.keys():\n image_hash = validated_data.pop(\"cover_photo\")\n photo = Photo.objects.filter(image_hash=image_hash).first()\n instance.cover_photo = photo\n instance.cover_face = photo.faces.filter(person__name=instance.name).first()\n instance.save()\n return instance\n return instance\n\n def delete(self, validated_data, id):\n person = Person.objects.filter(id=id).get()\n person.delete()\n\n\nclass AlbumPersonListSerializer(serializers.ModelSerializer):\n photo_count = serializers.SerializerMethodField()\n cover_photo_url = serializers.SerializerMethodField()\n\n class Meta:\n model = Person\n fields = (\n \"name\",\n \"photo_count\",\n \"cover_photo_url\",\n \"id\",\n )\n\n def get_photo_count(self, obj) -> int:\n return obj.filter(Q(person__is_null=False)).faces.count()\n\n def get_cover_photo_url(self, obj) -> str:\n first_face = obj.faces.filter(Q(person__is_null=False)).first()\n if first_face:\n return first_face.photo.thumbnail.square_thumbnail.url\n else:\n return None\n\n def get_face_photo_url(self, obj) -> str:\n first_face = obj.faces.filter(Q(person__is_null=False)).first()\n if first_face:\n return first_face.photo.image.url\n else:\n return None\n" }, { "path": "api/serializers/photo_metadata.py", "content": "\"\"\"\nSerializers for PhotoMetadata, MetadataFile, and MetadataEdit models.\n\nThese serializers provide:\n- Structured metadata access (replacing exif_json blob)\n- Edit history tracking\n- Backwards-compatible field names for existing API consumers\n\"\"\"\n\nfrom rest_framework import serializers\n\nfrom api.models import Photo\nfrom api.models.photo_metadata import MetadataEdit, MetadataFile, PhotoMetadata\n\n\nclass MetadataFileSerializer(serializers.ModelSerializer):\n \"\"\"Serializer for XMP sidecars and other metadata files.\"\"\"\n\n class Meta:\n model = MetadataFile\n fields = (\n \"id\",\n \"file_type\",\n \"source\",\n \"priority\",\n \"creator_software\",\n \"created_at\",\n \"updated_at\",\n )\n read_only_fields = (\"id\", \"created_at\", \"updated_at\")\n\n\nclass MetadataEditSerializer(serializers.ModelSerializer):\n \"\"\"Serializer for metadata edit history.\"\"\"\n\n user_name = serializers.SerializerMethodField()\n\n class Meta:\n model = MetadataEdit\n fields = (\n \"id\",\n \"field_name\",\n \"old_value\",\n \"new_value\",\n \"user\",\n \"user_name\",\n \"synced_to_file\",\n \"synced_at\",\n \"created_at\",\n )\n read_only_fields = fields\n\n def get_user_name(self, obj) -> str:\n if obj.user:\n return obj.user.username\n return \"Unknown\"\n\n\nclass PhotoMetadataSerializer(serializers.ModelSerializer):\n \"\"\"\n Full metadata serializer with all structured fields.\n \n Used for the detailed metadata view and editing.\n \"\"\"\n\n # Computed properties\n resolution = serializers.ReadOnlyField()\n megapixels = serializers.ReadOnlyField()\n has_location = serializers.ReadOnlyField()\n camera_display = serializers.ReadOnlyField()\n lens_display = serializers.ReadOnlyField()\n\n # Related data\n edit_history = serializers.SerializerMethodField()\n sidecar_files = serializers.SerializerMethodField()\n\n class Meta:\n model = PhotoMetadata\n fields = (\n \"id\",\n # Camera settings\n \"aperture\",\n \"shutter_speed\",\n \"shutter_speed_seconds\",\n \"iso\",\n \"focal_length\",\n \"focal_length_35mm\",\n \"exposure_compensation\",\n \"flash_fired\",\n \"metering_mode\",\n \"white_balance\",\n # Camera/lens info\n \"camera_make\",\n \"camera_model\",\n \"lens_make\",\n \"lens_model\",\n \"serial_number\",\n \"camera_display\",\n \"lens_display\",\n # Image properties\n \"width\",\n \"height\",\n \"orientation\",\n \"color_space\",\n \"bit_depth\",\n \"resolution\",\n \"megapixels\",\n # Timestamps\n \"date_taken\",\n \"date_taken_subsec\",\n \"date_modified\",\n \"timezone_offset\",\n # Location\n \"gps_latitude\",\n \"gps_longitude\",\n \"gps_altitude\",\n \"location_country\",\n \"location_state\",\n \"location_city\",\n \"location_address\",\n \"has_location\",\n # Content\n \"title\",\n \"caption\",\n \"keywords\",\n \"rating\",\n \"copyright\",\n \"creator\",\n # Tracking\n \"source\",\n \"version\",\n \"created_at\",\n \"updated_at\",\n # Related\n \"edit_history\",\n \"sidecar_files\",\n )\n read_only_fields = (\n \"id\",\n \"resolution\",\n \"megapixels\",\n \"has_location\",\n \"camera_display\",\n \"lens_display\",\n \"version\",\n \"created_at\",\n \"updated_at\",\n )\n\n def get_edit_history(self, obj) -> list:\n \"\"\"Get recent edit history for this photo.\"\"\"\n edits = MetadataEdit.objects.filter(photo=obj.photo).order_by(\"-created_at\")[:10]\n return MetadataEditSerializer(edits, many=True).data\n\n def get_sidecar_files(self, obj) -> list:\n \"\"\"Get sidecar files for this photo.\"\"\"\n files = MetadataFile.objects.filter(photo=obj.photo)\n return MetadataFileSerializer(files, many=True).data\n\n\nclass PhotoMetadataUpdateSerializer(serializers.ModelSerializer):\n \"\"\"\n Serializer for updating metadata with change tracking.\n \n Only allows editing specific fields and automatically\n creates MetadataEdit records for history.\n \"\"\"\n\n class Meta:\n model = PhotoMetadata\n fields = (\n # Editable fields\n \"title\",\n \"caption\",\n \"keywords\",\n \"rating\",\n \"copyright\",\n \"creator\",\n # Location (can be user-corrected)\n \"gps_latitude\",\n \"gps_longitude\",\n \"location_country\",\n \"location_state\",\n \"location_city\",\n \"location_address\",\n # Timestamp (can be user-corrected)\n \"date_taken\",\n \"timezone_offset\",\n )\n\n def update(self, instance, validated_data):\n \"\"\"Update metadata and create edit history records.\"\"\"\n user = self.context.get(\"request\").user if self.context.get(\"request\") else None\n \n for field_name, new_value in validated_data.items():\n old_value = getattr(instance, field_name)\n \n # Only track actual changes\n if old_value != new_value:\n # Create edit history record\n MetadataEdit.objects.create(\n photo=instance.photo,\n user=user,\n field_name=field_name,\n old_value=old_value,\n new_value=new_value,\n )\n \n # Update the field\n setattr(instance, field_name, new_value)\n \n # Update source to user_edit and increment version\n instance.source = PhotoMetadata.Source.USER_EDIT\n instance.version += 1\n instance.save()\n \n return instance\n\n\nclass PhotoMetadataSummarySerializer(serializers.Serializer):\n \"\"\"\n Lightweight metadata summary for photo lists.\n\n Returns key metadata fields without the full detail.\n \"\"\"\n\n # Camera info\n camera_display = serializers.CharField(allow_null=True)\n lens_display = serializers.CharField(allow_null=True)\n # Capture settings\n aperture = serializers.FloatField(allow_null=True)\n shutter_speed = serializers.CharField(allow_null=True)\n iso = serializers.IntegerField(allow_null=True)\n focal_length = serializers.FloatField(allow_null=True)\n focal_length_35mm = serializers.IntegerField(allow_null=True)\n # Image info\n resolution = serializers.CharField(allow_null=True)\n megapixels = serializers.FloatField(allow_null=True)\n # Date/location\n date_taken = serializers.DateTimeField(allow_null=True)\n has_location = serializers.BooleanField()\n # Content\n rating = serializers.IntegerField(allow_null=True)\n # Edit tracking\n source = serializers.CharField()\n version = serializers.IntegerField()\n has_edits = serializers.SerializerMethodField()\n\n def get_has_edits(self, obj) -> bool:\n \"\"\"Check if this photo has any metadata edits.\"\"\"\n return MetadataEdit.objects.filter(photo=obj.photo).exists()\n\n\ndef get_backwards_compatible_metadata(photo: Photo) -> dict:\n \"\"\"\n Generate backwards-compatible metadata dict from PhotoMetadata.\n \n This function returns metadata in the same format as the original\n Photo model fields for API backwards compatibility.\n \n Note: Metadata fields have been fully migrated to PhotoMetadata model.\n If no PhotoMetadata exists, return None/empty values.\n \"\"\"\n try:\n metadata = photo.metadata\n return {\n \"camera\": metadata.camera_display,\n \"lens\": metadata.lens_display,\n \"fstop\": metadata.aperture,\n \"focal_length\": metadata.focal_length,\n \"iso\": metadata.iso,\n \"shutter_speed\": metadata.shutter_speed,\n \"width\": metadata.width,\n \"height\": metadata.height,\n \"focalLength35Equivalent\": metadata.focal_length_35mm,\n \"digitalZoomRatio\": None, # Not stored in PhotoMetadata\n \"subjectDistance\": None, # Not stored in PhotoMetadata\n }\n except PhotoMetadata.DoesNotExist:\n # No PhotoMetadata exists - return None/empty values\n # Metadata will be populated on next photo scan\n return {\n \"camera\": None,\n \"lens\": None,\n \"fstop\": None,\n \"focal_length\": None,\n \"iso\": None,\n \"shutter_speed\": None,\n \"width\": 0,\n \"height\": 0,\n \"focalLength35Equivalent\": None,\n \"digitalZoomRatio\": None,\n \"subjectDistance\": None,\n }\n" }, { "path": "api/serializers/photos.py", "content": "import json\n\nfrom rest_framework import serializers\n\nfrom api.geocode.geocode import reverse_geocode\nfrom api.geocode import GEOCODE_VERSION\nfrom api import util\n\nfrom api.image_similarity import search_similar_image\nfrom api.models import AlbumDate, File, Photo\nfrom api.models.photo_metadata import PhotoMetadata\nfrom api.serializers.photo_metadata import PhotoMetadataSummarySerializer\nfrom api.serializers.simple import SimpleUserSerializer\n\n\nclass PhotoSummarySerializer(serializers.ModelSerializer):\n # UUID primary key\n id = serializers.UUIDField(read_only=True)\n # Content hash for deduplication/caching (legacy 'id' field for backwards compatibility)\n image_hash = serializers.CharField(read_only=True)\n dominantColor = serializers.SerializerMethodField()\n aspectRatio = serializers.SerializerMethodField()\n url = serializers.SerializerMethodField()\n location = serializers.SerializerMethodField()\n date = serializers.SerializerMethodField()\n birthTime = serializers.SerializerMethodField()\n video_length = serializers.SerializerMethodField()\n type = serializers.SerializerMethodField()\n owner = SimpleUserSerializer()\n # Stack information (can be multiple stacks)\n stacks = serializers.SerializerMethodField()\n # Flag indicating if this photo has a RAW file variant (PhotoPrism-like model)\n has_raw_variant = serializers.SerializerMethodField()\n\n class Meta:\n model = Photo\n fields = (\n \"id\",\n \"image_hash\",\n \"dominantColor\",\n \"url\",\n \"location\",\n \"date\",\n \"birthTime\",\n \"aspectRatio\",\n \"type\",\n \"video_length\",\n \"rating\",\n \"owner\",\n \"exif_gps_lat\",\n \"exif_gps_lon\",\n \"removed\",\n \"in_trashcan\",\n \"stacks\",\n \"has_raw_variant\",\n )\n\n # TODO: Rename this field to aspect_ratio\n def get_aspectRatio(self, obj) -> float:\n return obj.thumbnail.aspect_ratio\n\n # TODO: Remove this field in the future (kept for backwards compatibility)\n def get_url(self, obj) -> str:\n return obj.image_hash\n\n def get_location(self, obj) -> str:\n if (\n hasattr(obj, \"search_instance\")\n and obj.search_instance\n and obj.search_instance.search_location\n ):\n return obj.search_instance.search_location\n else:\n return \"\"\n\n def get_date(self, obj) -> str:\n if obj.exif_timestamp:\n return obj.exif_timestamp.isoformat()\n else:\n return \"\"\n\n def get_video_length(self, obj) -> int:\n if obj.video_length:\n return obj.video_length\n else:\n return \"\"\n\n # TODO: Remove this field in the future\n def get_birthTime(self, obj) -> str:\n if obj.exif_timestamp:\n return obj.exif_timestamp\n else:\n return \"\"\n\n def get_dominantColor(self, obj) -> str:\n if obj.thumbnail.dominant_color:\n dominant_color = obj.thumbnail.dominant_color[1:-1]\n return \"#%02x%02x%02x\" % tuple(map(int, dominant_color.split(\", \")))\n else:\n return \"\"\n\n def get_type(self, obj) -> str:\n if obj.video:\n return \"video\"\n # Use len() instead of .count() to leverage prefetched embedded_media\n if obj.main_file and len(obj.main_file.embedded_media.all()) > 0:\n return \"motion_photo\"\n return \"image\"\n\n def get_stacks(self, obj) -> list | None:\n \"\"\"Return stack info if photo is part of any stacks.\n\n Uses prefetched stacks data when available (from AlbumDateViewSet)\n to avoid N+1 queries. The prefetch filters by valid stack types and\n annotates photo_count_annotation so no extra queries are needed.\n \"\"\"\n # Use the prefetch cache (obj.stacks.all() won't re-query if prefetched)\n stacks = obj.stacks.all()\n if not stacks:\n return None\n\n from api.models.photo_stack import PhotoStack\n valid_stack_types = set(PhotoStack.VALID_STACK_TYPES + [\n PhotoStack.StackType.RAW_JPEG_PAIR,\n PhotoStack.StackType.LIVE_PHOTO,\n ])\n\n result = []\n for stack in stacks:\n # If stacks were prefetched with the type filter, all results are valid.\n # If not prefetched (called from another serializer context), filter in Python.\n if stack.stack_type not in valid_stack_types:\n continue\n is_primary = stack.primary_photo_id == obj.pk if stack.primary_photo_id else False\n # Use annotated count if available, otherwise fall back to DB query\n photo_count = getattr(stack, \"photo_count_annotation\", None)\n if photo_count is None:\n photo_count = stack.photos.count()\n result.append({\n \"id\": str(stack.id),\n \"type\": stack.stack_type,\n \"photo_count\": photo_count,\n \"is_primary\": is_primary,\n })\n\n return result or None\n\n def get_has_raw_variant(self, obj) -> bool:\n \"\"\"Check if this photo has a RAW file variant.\n\n Uses prefetched files when available to avoid N+1 queries.\n Returns True if any of the photo's files is a RAW file type.\n \"\"\"\n # Use prefetch cache if available (iterate in Python), otherwise query DB\n # File.RAW_FILE = 4\n if \"files\" in getattr(obj, \"_prefetched_objects_cache\", {}):\n return any(f.type == 4 for f in obj.files.all())\n return obj.files.filter(type=4).exists()\n\n\nclass GroupedPhotosSerializer(serializers.ModelSerializer):\n items = serializers.SerializerMethodField()\n date = serializers.SerializerMethodField()\n location = serializers.SerializerMethodField()\n\n class Meta:\n model = Photo\n fields = (\"date\", \"location\", \"items\")\n\n def get_date(self, obj) -> str:\n return obj.date\n\n def get_location(self, obj) -> str:\n return obj.location\n\n def get_items(self, obj) -> PhotoSummarySerializer(many=True):\n return PhotoSummarySerializer(obj.photos, many=True).data\n\n\nclass PhotoEditSerializer(serializers.ModelSerializer):\n class Meta:\n model = Photo\n fields = (\n \"image_hash\",\n \"hidden\",\n \"rating\",\n \"in_trashcan\",\n \"removed\",\n \"video\",\n \"exif_timestamp\",\n \"timestamp\",\n # Allow updating GPS location\n \"exif_gps_lat\",\n \"exif_gps_lon\",\n )\n\n def update(self, instance, validated_data):\n # photo can only update the following\n if \"exif_timestamp\" in validated_data:\n instance.timestamp = validated_data.pop(\"exif_timestamp\")\n instance.save()\n instance._extract_date_time_from_exif()\n\n # Update GPS location if provided\n lat = validated_data.pop(\"exif_gps_lat\", None)\n lon = validated_data.pop(\"exif_gps_lon\", None)\n\n if lat is not None and lon is not None:\n try:\n # Track old places to update album place relations\n old_album_places = instance._find_album_place()\n\n instance.exif_gps_lat = float(lat)\n instance.exif_gps_lon = float(lon)\n instance.save()\n\n # Reverse geocode and update geolocation/search location\n geocode_result = reverse_geocode(\n instance.exif_gps_lat, instance.exif_gps_lon\n )\n if geocode_result:\n geocode_result[\"_v\"] = GEOCODE_VERSION\n instance.geolocation_json = geocode_result\n\n # Update search location through PhotoSearch model\n from api.models.photo_search import PhotoSearch\n\n search_instance, _created = PhotoSearch.objects.get_or_create(\n photo=instance\n )\n search_instance.update_search_location(geocode_result)\n search_instance.save()\n\n # Update album place relations\n if old_album_places is not None:\n for old_album_place in old_album_places:\n old_album_place.photos.remove(instance)\n old_album_place.save()\n\n if \"features\" in geocode_result:\n for geolocation_level, feature in enumerate(\n geocode_result[\"features\"]\n ):\n if (\n \"text\" not in feature.keys()\n or str(feature[\"text\"]).isnumeric()\n ):\n continue\n album_place = api.models.album_place.get_album_place(\n feature[\"text\"], owner=instance.owner\n )\n if (\n album_place.photos.filter(\n image_hash=instance.image_hash\n ).count()\n == 0\n ):\n album_place.geolocation_level = (\n len(geocode_result[\"features\"]) - geolocation_level\n )\n album_place.photos.add(instance)\n album_place.save()\n\n instance.save()\n else:\n util.logger.warning(\n \"Reverse geocoding returned no result for provided coordinates\"\n )\n except Exception as e:\n util.logger.warning(e)\n util.logger.warning(\"Failed to update GPS location for photo\")\n return instance\n\n\nclass PhotoHashListSerializer(serializers.ModelSerializer):\n class Meta:\n model = Photo\n fields = (\"image_hash\", \"video\")\n\n\nclass PhotoDetailsSummarySerializer(serializers.ModelSerializer):\n photo_summary = serializers.SerializerMethodField()\n album_date_id = serializers.SerializerMethodField()\n processing = serializers.SerializerMethodField()\n\n class Meta:\n model = Photo\n fields = (\"photo_summary\", \"album_date_id\", \"processing\")\n\n def get_photo_summary(self, obj) -> PhotoSummarySerializer:\n return PhotoSummarySerializer(obj.get()).data\n\n def get_processing(self, obj) -> bool:\n return obj.get().thumbnail.aspect_ratio is None\n\n def get_album_date_id(self, obj) -> int:\n return (\n AlbumDate.objects.filter(photos__in=obj)\n .values_list(\"id\", flat=True)\n .first()\n )\n\n\nclass PhotoSerializer(serializers.ModelSerializer):\n square_thumbnail_url = serializers.SerializerMethodField()\n big_thumbnail_url = serializers.SerializerMethodField()\n small_square_thumbnail_url = serializers.SerializerMethodField()\n similar_photos = serializers.SerializerMethodField()\n captions_json = serializers.SerializerMethodField()\n search_captions = serializers.SerializerMethodField()\n search_location = serializers.SerializerMethodField()\n people = serializers.SerializerMethodField()\n shared_to = serializers.PrimaryKeyRelatedField(many=True, read_only=True)\n image_path = serializers.SerializerMethodField()\n owner = SimpleUserSerializer(many=False, read_only=True)\n embedded_media = serializers.SerializerMethodField()\n # File variants (RAW, JPEG, video for Live Photos, etc.)\n # PhotoPrism-like model where one Photo can have multiple file variants\n file_variants = serializers.SerializerMethodField()\n # Stack information (bursts, brackets, manual stacks) - can be multiple\n stacks = serializers.SerializerMethodField()\n # Structured metadata with edit history support\n metadata = serializers.SerializerMethodField()\n \n # Backwards-compatible fields from PhotoMetadata (for API compatibility)\n height = serializers.SerializerMethodField()\n width = serializers.SerializerMethodField()\n focal_length = serializers.SerializerMethodField()\n fstop = serializers.SerializerMethodField()\n iso = serializers.SerializerMethodField()\n shutter_speed = serializers.SerializerMethodField()\n lens = serializers.SerializerMethodField()\n camera = serializers.SerializerMethodField()\n focalLength35Equivalent = serializers.SerializerMethodField()\n digitalZoomRatio = serializers.SerializerMethodField()\n subjectDistance = serializers.SerializerMethodField()\n\n class Meta:\n model = Photo\n fields = (\n \"id\",\n \"exif_gps_lat\",\n \"exif_gps_lon\",\n \"exif_timestamp\",\n \"captions_json\",\n \"search_captions\",\n \"search_location\",\n \"big_thumbnail_url\",\n \"square_thumbnail_url\",\n \"small_square_thumbnail_url\",\n \"geolocation_json\",\n \"exif_json\",\n \"people\",\n \"image_hash\",\n \"image_path\",\n \"rating\",\n \"hidden\",\n \"public\",\n \"removed\",\n \"in_trashcan\",\n \"shared_to\",\n \"similar_photos\",\n \"video\",\n \"owner\",\n \"size\",\n \"height\",\n \"width\",\n \"focal_length\",\n \"fstop\",\n \"iso\",\n \"shutter_speed\",\n \"lens\",\n \"camera\",\n \"focalLength35Equivalent\",\n \"digitalZoomRatio\",\n \"subjectDistance\",\n \"embedded_media\",\n \"file_variants\",\n \"stacks\",\n \"metadata\",\n )\n \n def _get_metadata(self, obj) -> PhotoMetadata | None:\n \"\"\"Helper to get PhotoMetadata, with caching.\"\"\"\n if not hasattr(obj, '_cached_metadata'):\n try:\n obj._cached_metadata = obj.metadata\n except PhotoMetadata.DoesNotExist:\n obj._cached_metadata = None\n return obj._cached_metadata\n \n def get_height(self, obj) -> int:\n metadata = self._get_metadata(obj)\n return metadata.height if metadata else 0\n \n def get_width(self, obj) -> int:\n metadata = self._get_metadata(obj)\n return metadata.width if metadata else 0\n \n def get_focal_length(self, obj) -> float | None:\n metadata = self._get_metadata(obj)\n return metadata.focal_length if metadata else None\n \n def get_fstop(self, obj) -> float | None:\n metadata = self._get_metadata(obj)\n return metadata.aperture if metadata else None\n \n def get_iso(self, obj) -> int | None:\n metadata = self._get_metadata(obj)\n return metadata.iso if metadata else None\n \n def get_shutter_speed(self, obj) -> str | None:\n metadata = self._get_metadata(obj)\n return metadata.shutter_speed if metadata else None\n \n def get_lens(self, obj) -> str | None:\n metadata = self._get_metadata(obj)\n return metadata.lens_display if metadata else None\n \n def get_camera(self, obj) -> str | None:\n metadata = self._get_metadata(obj)\n return metadata.camera_display if metadata else None\n \n def get_focalLength35Equivalent(self, obj) -> int | None:\n metadata = self._get_metadata(obj)\n return metadata.focal_length_35mm if metadata else None\n \n def get_digitalZoomRatio(self, obj) -> float | None:\n # Not stored in PhotoMetadata (rarely used field)\n return None\n \n def get_subjectDistance(self, obj) -> float | None:\n # Not stored in PhotoMetadata (rarely used field)\n return None\n\n def get_similar_photos(self, obj) -> list:\n res = search_similar_image(obj.owner, obj, threshold=90)\n arr = []\n if len(res) > 0:\n [arr.append(e) for e in res[\"result\"]]\n photos = Photo.objects.filter(image_hash__in=arr).all()\n res = []\n for photo in photos:\n type = \"image\"\n if photo.video:\n type = \"video\"\n res.append({\"image_hash\": photo.image_hash, \"type\": type})\n return res\n else:\n return []\n\n def get_captions_json(self, obj) -> dict:\n if (\n hasattr(obj, \"caption_instance\")\n and obj.caption_instance\n and obj.caption_instance.captions_json\n and len(obj.caption_instance.captions_json) > 0\n ):\n return obj.caption_instance.captions_json\n else:\n emptyArray = {\n \"im2txt\": \"\",\n \"places365\": {\"attributes\": [], \"categories\": [], \"environment\": []},\n }\n return emptyArray\n\n def get_search_captions(self, obj) -> str:\n if hasattr(obj, \"search_instance\") and obj.search_instance:\n return obj.search_instance.search_captions or \"\"\n return \"\"\n\n def get_search_location(self, obj) -> str:\n if hasattr(obj, \"search_instance\") and obj.search_instance:\n return obj.search_instance.search_location or \"\"\n return \"\"\n\n def get_image_path(self, obj) -> list[str]:\n try:\n paths = []\n for file in obj.files.all():\n paths.append(file.path)\n return paths\n except Exception:\n return [\"Missing\"]\n\n def get_square_thumbnail_url(self, obj) -> str:\n return (\n obj.thumbnail.square_thumbnail.url if obj.thumbnail.square_thumbnail else \"\"\n )\n\n def get_small_square_thumbnail_url(self, obj) -> str:\n return (\n obj.thumbnail.square_thumbnail_small.url\n if obj.thumbnail.square_thumbnail_small\n else \"\"\n )\n\n def get_big_thumbnail_url(self, obj) -> str:\n return obj.thumbnail.thumbnail_big.url if obj.thumbnail.thumbnail_big else \"\"\n\n def get_geolocation(self, obj) -> dict:\n if obj.geolocation_json:\n return json.loads(obj.geolocation_json)\n else:\n return None\n\n def get_people(self, obj) -> list:\n return [\n {\n \"name\": (\n f.person.name\n if f.person\n else (\n f.cluster_person.name\n if f.cluster_person\n else (\n f.classification_person.name\n if f.classification_person\n else \"\"\n )\n )\n ),\n \"type\": (\n \"user\"\n if f.person\n else (\n \"cluster\"\n if f.cluster_person\n else (\"classification\" if f.classification_person else \"\")\n )\n ),\n \"probability\": (\n 1\n if f.person\n else (\n f.cluster_probability\n if f.cluster_person\n else (\n f.classification_probability\n if f.classification_person\n else 0\n )\n )\n ),\n \"location\": {\n \"top\": f.location_top,\n \"bottom\": f.location_bottom,\n \"left\": f.location_left,\n \"right\": f.location_right,\n },\n \"face_url\": f.image.url,\n \"face_id\": f.id,\n }\n for f in obj.faces.all()\n ]\n\n def get_embedded_media(self, obj: Photo) -> list[dict]:\n def serialize_file(file):\n return {\n \"id\": file.hash,\n \"type\": \"video\" if file.type == File.VIDEO else \"image\",\n }\n\n if obj.main_file is None:\n return []\n embedded_media = obj.main_file.embedded_media.all()\n if len(embedded_media) == 0:\n return []\n return list(\n map(\n serialize_file, embedded_media.filter(type__in=[File.VIDEO, File.IMAGE])\n )\n )\n\n def get_metadata(self, obj: Photo) -> dict | None:\n \"\"\"\n Return structured metadata from PhotoMetadata if available.\n \n This provides:\n - Normalized field names (aperture, iso, shutter_speed, etc.)\n - Computed display strings (camera_display, lens_display)\n - Resolution and megapixel info\n - Edit tracking (version, source, has_edits)\n \n Falls back to None if PhotoMetadata doesn't exist (backwards compatible).\n \"\"\"\n try:\n metadata = obj.metadata\n return PhotoMetadataSummarySerializer(metadata).data\n except PhotoMetadata.DoesNotExist:\n return None\n\n def get_file_variants(self, obj: Photo) -> list | None:\n \"\"\"Return file variants for this photo (RAW, JPEG, video for Live Photos, etc.).\n \n This implements the PhotoPrism-like model where one Photo can have multiple\n file variants representing the same capture moment.\n \"\"\"\n from api.models.file import File\n \n files = obj.files.all()\n if files.count() <= 1:\n # Only main file, no additional variants\n return None\n \n variants = []\n for f in files:\n # Determine file type label\n file_type_map = {\n File.IMAGE: \"image\",\n File.VIDEO: \"video\",\n File.RAW_FILE: \"raw\",\n File.METADATA_FILE: \"metadata\",\n File.UNKNOWN: \"unknown\",\n }\n file_type = file_type_map.get(f.type, \"unknown\")\n \n # Check if this is the main file\n is_main = obj.main_file_id == f.hash if obj.main_file_id else False\n \n variants.append({\n \"hash\": f.hash,\n \"path\": f.path,\n \"type\": file_type,\n \"type_id\": f.type,\n \"is_main\": is_main,\n \"filename\": f.path.split(\"/\")[-1] if f.path else None,\n })\n \n return variants\n\n def get_stacks(self, obj: Photo) -> list | None:\n \"\"\"Return detailed stack info for photo detail view (supports multiple stacks).\"\"\"\n from api.models.photo_stack import PhotoStack\n # Use model-defined valid stack types, plus deprecated types for backwards compatibility\n valid_stack_types = PhotoStack.VALID_STACK_TYPES + [\n PhotoStack.StackType.RAW_JPEG_PAIR,\n PhotoStack.StackType.LIVE_PHOTO,\n ]\n stacks = obj.stacks.filter(stack_type__in=valid_stack_types)\n if not stacks.exists():\n return None\n \n result = []\n for stack in stacks:\n is_primary = stack.primary_photo_id == obj.pk if stack.primary_photo_id else False\n \n # Get all photos in the stack for the detail view\n stack_photos = []\n for photo in stack.photos.select_related(\"thumbnail\").all():\n # Get width/height from PhotoMetadata\n try:\n photo_metadata = photo.metadata\n photo_width = photo_metadata.width or 0\n photo_height = photo_metadata.height or 0\n except PhotoMetadata.DoesNotExist:\n photo_width = 0\n photo_height = 0\n \n stack_photos.append({\n \"id\": str(photo.id),\n \"image_hash\": photo.image_hash,\n \"is_primary\": photo.pk == stack.primary_photo_id,\n \"thumbnail_url\": (\n f\"/media/square_thumbnails_small/{photo.image_hash}\"\n if hasattr(photo, \"thumbnail\") and photo.thumbnail and photo.thumbnail.square_thumbnail_small \n else None\n ),\n \"size\": photo.size,\n \"width\": photo_width,\n \"height\": photo_height,\n })\n \n result.append({\n \"id\": str(stack.id),\n \"type\": stack.stack_type,\n \"type_display\": stack.get_stack_type_display(),\n \"photo_count\": len(stack_photos),\n \"is_primary\": is_primary,\n \"photos\": stack_photos,\n })\n \n return result\n\n\nclass SharedFromMePhotoThroughSerializer(serializers.ModelSerializer):\n photo = serializers.SerializerMethodField()\n user = SimpleUserSerializer(many=False, read_only=True)\n\n class Meta:\n model = Photo.shared_to.through\n fields = (\"user_id\", \"user\", \"photo\")\n\n def get_photo(self, obj) -> PhotoSummarySerializer:\n return PhotoSummarySerializer(obj.photo).data\n\n\nclass PublicPhotoDetailSerializer(serializers.ModelSerializer):\n \"\"\"Serializer for photo details in public albums.\n \n Conditionally includes metadata based on sharing settings passed in context.\n Context must include 'sharing_settings' dict with keys:\n - share_location: bool\n - share_camera_info: bool\n - share_timestamps: bool\n - share_captions: bool\n - share_faces: bool\n \"\"\"\n \n # Always included\n square_thumbnail_url = serializers.SerializerMethodField()\n big_thumbnail_url = serializers.SerializerMethodField()\n small_square_thumbnail_url = serializers.SerializerMethodField()\n video = serializers.BooleanField(read_only=True)\n image_hash = serializers.CharField(read_only=True)\n \n # Conditionally included based on sharing settings\n exif_timestamp = serializers.SerializerMethodField()\n exif_gps_lat = serializers.SerializerMethodField()\n exif_gps_lon = serializers.SerializerMethodField()\n geolocation_json = serializers.SerializerMethodField()\n search_location = serializers.SerializerMethodField()\n \n # Camera info\n camera = serializers.SerializerMethodField()\n lens = serializers.SerializerMethodField()\n focal_length = serializers.SerializerMethodField()\n fstop = serializers.SerializerMethodField()\n iso = serializers.SerializerMethodField()\n shutter_speed = serializers.SerializerMethodField()\n width = serializers.SerializerMethodField()\n height = serializers.SerializerMethodField()\n \n # Captions\n search_captions = serializers.SerializerMethodField()\n captions_json = serializers.SerializerMethodField()\n \n # People/faces\n people = serializers.SerializerMethodField()\n\n class Meta:\n model = Photo\n fields = (\n \"image_hash\",\n \"video\",\n \"square_thumbnail_url\",\n \"big_thumbnail_url\",\n \"small_square_thumbnail_url\",\n \"exif_timestamp\",\n \"exif_gps_lat\",\n \"exif_gps_lon\",\n \"geolocation_json\",\n \"search_location\",\n \"camera\",\n \"lens\",\n \"focal_length\",\n \"fstop\",\n \"iso\",\n \"shutter_speed\",\n \"width\",\n \"height\",\n \"search_captions\",\n \"captions_json\",\n \"people\",\n )\n\n def _get_sharing_settings(self) -> dict:\n \"\"\"Get sharing settings from context.\"\"\"\n return self.context.get('sharing_settings', {})\n \n def _get_metadata(self, obj) -> PhotoMetadata | None:\n \"\"\"Helper to get PhotoMetadata.\"\"\"\n if not hasattr(obj, '_cached_metadata'):\n try:\n obj._cached_metadata = obj.metadata\n except PhotoMetadata.DoesNotExist:\n obj._cached_metadata = None\n return obj._cached_metadata\n\n # Always available\n def get_square_thumbnail_url(self, obj) -> str:\n return obj.thumbnail.square_thumbnail.url if obj.thumbnail and obj.thumbnail.square_thumbnail else \"\"\n\n def get_small_square_thumbnail_url(self, obj) -> str:\n return obj.thumbnail.square_thumbnail_small.url if obj.thumbnail and obj.thumbnail.square_thumbnail_small else \"\"\n\n def get_big_thumbnail_url(self, obj) -> str:\n return obj.thumbnail.thumbnail_big.url if obj.thumbnail and obj.thumbnail.thumbnail_big else \"\"\n\n # Timestamp - conditional\n def get_exif_timestamp(self, obj):\n if self._get_sharing_settings().get('share_timestamps', False):\n return obj.exif_timestamp\n return None\n\n # Location - conditional\n def get_exif_gps_lat(self, obj):\n if self._get_sharing_settings().get('share_location', False):\n return obj.exif_gps_lat\n return None\n\n def get_exif_gps_lon(self, obj):\n if self._get_sharing_settings().get('share_location', False):\n return obj.exif_gps_lon\n return None\n\n def get_geolocation_json(self, obj):\n if self._get_sharing_settings().get('share_location', False):\n return obj.geolocation_json\n return None\n\n def get_search_location(self, obj) -> str:\n if self._get_sharing_settings().get('share_location', False):\n if hasattr(obj, \"search_instance\") and obj.search_instance:\n return obj.search_instance.search_location or \"\"\n return \"\"\n\n # Camera info - conditional\n def get_camera(self, obj) -> str | None:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.camera_display if metadata else None\n return None\n\n def get_lens(self, obj) -> str | None:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.lens_display if metadata else None\n return None\n\n def get_focal_length(self, obj) -> float | None:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.focal_length if metadata else None\n return None\n\n def get_fstop(self, obj) -> float | None:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.aperture if metadata else None\n return None\n\n def get_iso(self, obj) -> int | None:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.iso if metadata else None\n return None\n\n def get_shutter_speed(self, obj) -> str | None:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.shutter_speed if metadata else None\n return None\n\n def get_width(self, obj) -> int:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.width if metadata else 0\n return 0\n\n def get_height(self, obj) -> int:\n if self._get_sharing_settings().get('share_camera_info', False):\n metadata = self._get_metadata(obj)\n return metadata.height if metadata else 0\n return 0\n\n # Captions - conditional\n def get_search_captions(self, obj) -> str:\n if self._get_sharing_settings().get('share_captions', False):\n if hasattr(obj, \"search_instance\") and obj.search_instance:\n return obj.search_instance.search_captions or \"\"\n return \"\"\n\n def get_captions_json(self, obj) -> dict:\n if self._get_sharing_settings().get('share_captions', False):\n if (\n hasattr(obj, \"caption_instance\")\n and obj.caption_instance\n and obj.caption_instance.captions_json\n and len(obj.caption_instance.captions_json) > 0\n ):\n return obj.caption_instance.captions_json\n return {\"im2txt\": \"\", \"places365\": {\"attributes\": [], \"categories\": [], \"environment\": []}}\n\n # People/faces - conditional\n def get_people(self, obj) -> list:\n if not self._get_sharing_settings().get('share_faces', False):\n return []\n \n return [\n {\n \"name\": (\n f.person.name\n if f.person\n else (\n f.cluster_person.name\n if f.cluster_person\n else \"Unknown\"\n )\n ),\n \"face_url\": f.image.url if f.image else None,\n \"face_id\": f.id,\n }\n for f in obj.faces.all()\n if f.person or f.cluster_person\n ]\n" }, { "path": "api/serializers/simple.py", "content": "from rest_framework import serializers\n\nfrom api.models import Photo, User\n\n\nclass PhotoSuperSimpleSerializer(serializers.ModelSerializer):\n class Meta:\n model = Photo\n fields = (\"image_hash\", \"rating\", \"hidden\", \"exif_timestamp\", \"public\", \"video\")\n\n\nclass PhotoSimpleSerializer(serializers.ModelSerializer):\n square_thumbnail = serializers.SerializerMethodField()\n\n class Meta:\n model = Photo\n fields = (\n \"square_thumbnail\",\n \"image_hash\",\n \"exif_timestamp\",\n \"exif_gps_lat\",\n \"exif_gps_lon\",\n \"rating\",\n \"geolocation_json\",\n \"public\",\n \"video\",\n )\n\n def get_square_thumbnail(self, obj) -> str:\n return (\n obj.thumbnail.square_thumbnail.url\n if obj.thumbnail and obj.thumbnail.square_thumbnail\n else \"\"\n )\n\n\nclass SimpleUserSerializer(serializers.ModelSerializer):\n class Meta:\n model = User\n fields = (\n \"id\",\n \"username\",\n \"first_name\",\n \"last_name\",\n )\n" }, { "path": "api/serializers/user.py", "content": "import os\n\nfrom django.conf import settings\nfrom django.contrib.auth import get_user_model\nfrom django.db.models import Q\nfrom django_q.tasks import Chain\nfrom rest_framework import serializers\nfrom rest_framework.exceptions import ValidationError\n\nfrom api.batch_jobs import batch_calculate_clip_embedding\nfrom api.ml_models import do_all_models_exist, download_models\nfrom api.models import Photo, User\nfrom api.serializers.simple import PhotoSuperSimpleSerializer\nfrom api.util import is_valid_path, logger\n\n\nclass UserSerializer(serializers.ModelSerializer):\n public_photo_count = serializers.SerializerMethodField()\n public_photo_samples = serializers.SerializerMethodField()\n photo_count = serializers.SerializerMethodField()\n avatar_url = serializers.SerializerMethodField()\n\n class Meta:\n model = User\n extra_kwargs = {\n \"password\": {\"write_only\": True},\n \"first_name\": {\"required\": False},\n \"last_name\": {\"required\": False},\n \"scan_directory\": {\"required\": False},\n \"confidence\": {\"required\": False},\n \"confidence_person\": {\"required\": False},\n \"semantic_search_topk\": {\"required\": False},\n \"nextcloud_server_address\": {\"required\": False},\n \"nextcloud_username\": {\"required\": False},\n \"nextcloud_scan_directory\": {\"required\": False},\n \"nextcloud_app_password\": {\"write_only\": True},\n \"favorite_min_rating\": {\"required\": False},\n \"save_metadata_to_disk\": {\"required\": False},\n \"save_face_tags_to_disk\": {\"required\": False},\n \"text_alignment\": {\"required\": False},\n \"header_size\": {\"required\": False},\n \"skip_raw_files\": {\"required\": False},\n \"stack_raw_jpeg\": {\"required\": False},\n \"slideshow_interval\": {\"required\": False},\n \"duplicate_sensitivity\": {\"required\": False},\n \"duplicate_clear_existing\": {\"required\": False},\n }\n fields = (\n \"id\",\n \"username\",\n \"email\",\n \"scan_directory\",\n \"confidence\",\n \"confidence_person\",\n \"transcode_videos\",\n \"semantic_search_topk\",\n \"first_name\",\n \"public_photo_samples\",\n \"last_name\",\n \"public_photo_count\",\n \"date_joined\",\n \"password\",\n \"avatar\",\n \"is_superuser\",\n \"photo_count\",\n \"nextcloud_server_address\",\n \"nextcloud_username\",\n \"nextcloud_app_password\",\n \"nextcloud_scan_directory\",\n \"avatar_url\",\n \"favorite_min_rating\",\n \"image_scale\",\n \"text_alignment\",\n \"header_size\",\n \"save_metadata_to_disk\",\n \"save_face_tags_to_disk\",\n \"datetime_rules\",\n \"burst_detection_rules\",\n \"llm_settings\",\n \"default_timezone\",\n \"public_sharing\",\n \"public_sharing_defaults\",\n \"face_recognition_model\",\n \"min_cluster_size\",\n \"confidence_unknown_face\",\n \"min_samples\",\n \"cluster_selection_epsilon\",\n \"skip_raw_files\",\n \"stack_raw_jpeg\",\n \"slideshow_interval\",\n \"duplicate_sensitivity\",\n \"duplicate_clear_existing\",\n )\n\n def validate_nextcloud_app_password(self, value):\n return value\n\n def create(self, validated_data):\n if \"scan_directory\" in validated_data.keys():\n if (\n not self.context[\"request\"].user.is_superuser\n or validated_data[\"scan_directory\"] == \"initial\"\n ):\n validated_data.pop(\"scan_directory\")\n # make sure username is always lowercase\n if \"username\" in validated_data.keys():\n validated_data[\"username\"] = validated_data[\"username\"].lower()\n if \"is_superuser\" in validated_data.keys():\n is_superuser = validated_data.pop(\"is_superuser\")\n if is_superuser and self.context[\"request\"].user.is_authenticated and self.context[\"request\"].user.is_superuser:\n user = User.objects.create_superuser(**validated_data)\n else:\n user = User.objects.create_user(**validated_data)\n else:\n user = User.objects.create_user(**validated_data)\n logger.info(f\"Created user {user.id}\")\n return user\n\n def update(self, instance, validated_data):\n # user can only update the following\n if \"password\" in validated_data:\n password = validated_data.pop(\"password\")\n if password != \"\" and not settings.DEMO_SITE:\n instance.set_password(password)\n if \"avatar\" in validated_data:\n instance.avatar = validated_data.pop(\"avatar\")\n instance.save()\n if \"email\" in validated_data:\n instance.email = validated_data.pop(\"email\")\n instance.save()\n if \"first_name\" in validated_data:\n instance.first_name = validated_data.pop(\"first_name\")\n instance.save()\n if \"last_name\" in validated_data:\n instance.last_name = validated_data.pop(\"last_name\")\n instance.save()\n if \"transcode_videos\" in validated_data:\n instance.transcode_videos = validated_data.pop(\"transcode_videos\")\n instance.save()\n if \"nextcloud_server_address\" in validated_data:\n instance.nextcloud_server_address = validated_data.pop(\n \"nextcloud_server_address\"\n )\n instance.save()\n if \"nextcloud_username\" in validated_data:\n instance.nextcloud_username = validated_data.pop(\"nextcloud_username\")\n instance.save()\n if \"nextcloud_app_password\" in validated_data:\n instance.nextcloud_app_password = validated_data.pop(\n \"nextcloud_app_password\"\n )\n instance.save()\n if \"nextcloud_scan_directory\" in validated_data:\n instance.nextcloud_scan_directory = validated_data.pop(\n \"nextcloud_scan_directory\"\n )\n instance.save()\n if \"confidence\" in validated_data:\n instance.confidence = validated_data.pop(\"confidence\")\n instance.save()\n logger.info(f\"Updated confidence for user {instance.confidence}\")\n if \"confidence_person\" in validated_data:\n instance.confidence_person = validated_data.pop(\"confidence_person\")\n instance.save()\n logger.info(\n f\"Updated person album confidence for user {instance.confidence_person}\"\n )\n if \"semantic_search_topk\" in validated_data:\n new_semantic_search_topk = validated_data.pop(\"semantic_search_topk\")\n\n if instance.semantic_search_topk == 0 and new_semantic_search_topk > 0:\n chain = Chain()\n if not do_all_models_exist():\n chain.append(download_models, User.objects.get(id=instance.id))\n chain.append(\n batch_calculate_clip_embedding, User.objects.get(id=instance.id)\n )\n chain.run()\n\n instance.semantic_search_topk = new_semantic_search_topk\n instance.save()\n logger.info(\n f\"Updated semantic_search_topk for user {instance.semantic_search_topk}\"\n )\n if \"favorite_min_rating\" in validated_data:\n new_favorite_min_rating = validated_data.pop(\"favorite_min_rating\")\n instance.favorite_min_rating = new_favorite_min_rating\n instance.save()\n logger.info(\n f\"Updated favorite_min_rating for user {instance.favorite_min_rating}\"\n )\n if \"save_metadata_to_disk\" in validated_data:\n instance.save_metadata_to_disk = validated_data.pop(\"save_metadata_to_disk\")\n instance.save()\n logger.info(\n f\"Updated save_metadata_to_disk for user {instance.save_metadata_to_disk}\"\n )\n if \"save_face_tags_to_disk\" in validated_data:\n instance.save_face_tags_to_disk = validated_data.pop(\n \"save_face_tags_to_disk\"\n )\n instance.save()\n logger.info(\n f\"Updated save_face_tags_to_disk to {instance.save_face_tags_to_disk} for user {instance.username}\"\n )\n if \"image_scale\" in validated_data:\n new_image_scale = validated_data.pop(\"image_scale\")\n instance.image_scale = new_image_scale\n instance.save()\n logger.info(f\"Updated image_scale for user {instance.image_scale}\")\n if \"text_alignment\" in validated_data:\n new_text_alignment = validated_data.pop(\"text_alignment\")\n instance.text_alignment = new_text_alignment\n instance.save()\n logger.info(f\"Updated text_alignment for user {instance.text_alignment}\")\n if \"header_size\" in validated_data:\n new_header_size = validated_data.pop(\"header_size\")\n instance.header_size = new_header_size\n instance.save()\n logger.info(f\"Updated header_size for user {instance.header_size}\")\n if \"datetime_rules\" in validated_data:\n new_datetime_rules = validated_data.pop(\"datetime_rules\")\n instance.datetime_rules = new_datetime_rules\n instance.save()\n logger.info(f\"Updated datetime_rules for user {instance.datetime_rules}\")\n if \"default_timezone\" in validated_data:\n new_default_timezone = validated_data.pop(\"default_timezone\")\n instance.default_timezone = new_default_timezone\n instance.save()\n logger.info(\n f\"Updated default_timezone for user {instance.default_timezone}\"\n )\n if \"public_sharing\" in validated_data:\n instance.public_sharing = validated_data.pop(\"public_sharing\")\n instance.save()\n if \"face_recognition_model\" in validated_data:\n instance.face_recognition_model = validated_data.pop(\n \"face_recognition_model\"\n )\n instance.save()\n if \"min_cluster_size\" in validated_data:\n instance.min_cluster_size = validated_data.pop(\"min_cluster_size\")\n instance.save()\n if \"confidence_unknown_face\" in validated_data:\n instance.confidence_unknown_face = validated_data.pop(\n \"confidence_unknown_face\"\n )\n instance.save()\n if \"min_samples\" in validated_data:\n instance.min_samples = validated_data.pop(\"min_samples\")\n instance.save()\n if \"cluster_selection_epsilon\" in validated_data:\n instance.cluster_selection_epsilon = validated_data.pop(\n \"cluster_selection_epsilon\"\n )\n instance.save()\n if \"llm_settings\" in validated_data:\n instance.llm_settings = validated_data.pop(\"llm_settings\")\n instance.save()\n if \"skip_raw_files\" in validated_data:\n instance.skip_raw_files = validated_data.pop(\"skip_raw_files\")\n instance.save()\n logger.info(\n f\"Updated skip_raw_files to {instance.skip_raw_files} for user {instance.username}\"\n )\n if \"stack_raw_jpeg\" in validated_data:\n instance.stack_raw_jpeg = validated_data.pop(\"stack_raw_jpeg\")\n instance.save()\n logger.info(\n f\"Updated stack_raw_jpeg to {instance.stack_raw_jpeg} for user {instance.username}\"\n )\n if \"slideshow_interval\" in validated_data:\n instance.slideshow_interval = validated_data.pop(\"slideshow_interval\")\n instance.save()\n logger.info(\n f\"Updated slideshow_interval to {instance.slideshow_interval} for user {instance.username}\"\n )\n if \"duplicate_sensitivity\" in validated_data:\n instance.duplicate_sensitivity = validated_data.pop(\"duplicate_sensitivity\")\n instance.save()\n logger.info(\n f\"Updated duplicate_sensitivity to {instance.duplicate_sensitivity} for user {instance.username}\"\n )\n if \"duplicate_clear_existing\" in validated_data:\n instance.duplicate_clear_existing = validated_data.pop(\n \"duplicate_clear_existing\"\n )\n instance.save()\n logger.info(\n f\"Updated duplicate_clear_existing to {instance.duplicate_clear_existing} for user {instance.username}\"\n )\n\n return instance\n\n def get_photo_count(self, obj) -> int:\n return Photo.objects.filter(owner=obj).count()\n\n def get_public_photo_count(self, obj) -> int:\n return Photo.objects.filter(Q(owner=obj) & Q(public=True)).count()\n\n def get_public_photo_samples(self, obj) -> PhotoSuperSimpleSerializer(many=True):\n return PhotoSuperSimpleSerializer(\n Photo.objects.filter(Q(owner=obj) & Q(public=True))[:10], many=True\n ).data\n\n def get_avatar_url(self, obj) -> str or None:\n try:\n return obj.avatar.url\n except Exception:\n return None\n\n\nclass PublicUserSerializer(serializers.ModelSerializer):\n public_photo_count = serializers.SerializerMethodField()\n public_photo_samples = serializers.SerializerMethodField()\n avatar_url = serializers.SerializerMethodField()\n\n class Meta:\n model = User\n fields = (\n \"id\",\n \"avatar_url\",\n \"username\",\n \"first_name\",\n \"last_name\",\n \"public_photo_count\",\n \"public_photo_samples\",\n )\n\n def get_public_photo_count(self, obj) -> int:\n return Photo.objects.filter(Q(owner=obj) & Q(public=True)).count()\n\n def get_public_photo_samples(self, obj) -> PhotoSuperSimpleSerializer(many=True):\n return PhotoSuperSimpleSerializer(\n Photo.objects.filter(Q(owner=obj) & Q(public=True))[:10], many=True\n ).data\n\n def get_avatar_url(self, obj) -> str or None:\n try:\n return obj.avatar.url\n except ValueError:\n return None\n\n\nclass SignupUserSerializer(serializers.ModelSerializer):\n class Meta:\n model = User\n extra_kwargs = {\n \"username\": {\"required\": True},\n \"password\": {\n \"write_only\": True,\n \"required\": True,\n \"min_length\": 3, # configurable min password length?\n },\n \"email\": {\"required\": True},\n \"first_name\": {\"required\": True},\n \"last_name\": {\"required\": True},\n \"is_superuser\": {\"write_only\": True},\n }\n fields = (\n \"username\",\n \"password\",\n \"email\",\n \"first_name\",\n \"last_name\",\n \"is_superuser\",\n )\n\n def create(self, validated_data):\n should_be_superuser = User.objects.filter(is_superuser=True).count() == 0\n user = super().create(validated_data)\n user.set_password(validated_data.pop(\"password\"))\n user.is_staff = should_be_superuser\n user.is_superuser = should_be_superuser\n user.save()\n return user\n\n\nclass DeleteUserSerializer(serializers.ModelSerializer):\n class Meta:\n model = get_user_model()\n fields = \"__all__\"\n\n\nclass ManageUserSerializer(serializers.ModelSerializer):\n photo_count = serializers.SerializerMethodField()\n\n class Meta:\n model = get_user_model()\n fields = (\n \"username\",\n \"scan_directory\",\n \"skip_raw_files\",\n \"stack_raw_jpeg\",\n \"confidence\",\n \"semantic_search_topk\",\n \"last_login\",\n \"date_joined\",\n \"photo_count\",\n \"id\",\n \"favorite_min_rating\",\n \"image_scale\",\n \"save_metadata_to_disk\",\n \"email\",\n \"first_name\",\n \"last_name\",\n \"password\",\n )\n extra_kwargs = {\n \"password\": {\"write_only\": True},\n \"scan_directory\": {\"required\": False},\n \"skip_raw_files\": {\"required\": False},\n \"stack_raw_jpeg\": {\"required\": False},\n }\n\n def get_photo_count(self, obj) -> int:\n return Photo.objects.filter(owner=obj).count()\n\n def update(self, instance: User, validated_data):\n if \"password\" in validated_data:\n password = validated_data.pop(\"password\")\n if password != \"\" and not settings.DEMO_SITE:\n instance.set_password(password)\n\n if \"scan_directory\" in validated_data:\n new_scan_directory = validated_data.pop(\"scan_directory\")\n\n if new_scan_directory: # Ensure it's not an empty string\n abs_new_scan_directory = os.path.abspath(new_scan_directory)\n\n if not is_valid_path(abs_new_scan_directory, settings.DATA_ROOT):\n raise ValidationError(\n \"Scan directory must be inside the data root.\"\n )\n\n if os.path.exists(abs_new_scan_directory):\n instance.scan_directory = abs_new_scan_directory\n logger.info(\n f\"Updated scan directory for user {instance.scan_directory}\"\n )\n else:\n raise ValidationError(\"Scan directory does not exist\")\n if \"skip_raw_files\" in validated_data:\n instance.skip_raw_files = validated_data.pop(\"skip_raw_files\")\n if \"stack_raw_jpeg\" in validated_data:\n instance.stack_raw_jpeg = validated_data.pop(\"stack_raw_jpeg\")\n\n if \"username\" in validated_data:\n username = validated_data.pop(\"username\")\n if username != \"\":\n other_user = User.objects.filter(username=username).first()\n if other_user is not None and other_user != instance:\n raise ValidationError(\"User name is already taken\")\n\n instance.username = username\n\n if \"email\" in validated_data:\n email = validated_data.pop(\"email\")\n instance.email = email\n\n if \"first_name\" in validated_data:\n first_name = validated_data.pop(\"first_name\")\n instance.first_name = first_name\n\n if \"last_name\" in validated_data:\n last_name = validated_data.pop(\"last_name\")\n instance.last_name = last_name\n\n instance.save()\n return instance\n" }, { "path": "api/services.py", "content": "import platform\nimport subprocess\nimport time\nfrom datetime import timedelta\n\nimport requests\nfrom django.db.models import Q\nfrom django.utils import timezone\n\nfrom api.models import Photo\nfrom api.util import logger\n\n# Track services that should not be restarted due to system incompatibility\nINCOMPATIBLE_SERVICES = set()\n\n# CPU features required for different services\nSERVICE_CPU_REQUIREMENTS = {\n \"llm\": {\n \"required\": [\"avx\", \"sse4_2\"], # Essential for llama.cpp\n \"recommended\": [\"avx2\", \"fma\", \"f16c\"], # Improve performance\n }\n}\n\n# Define all the services that can be started, with their respective ports\nSERVICES = {\n \"image_similarity\": 8002,\n \"thumbnail\": 8003,\n \"face_recognition\": 8005,\n \"clip_embeddings\": 8006,\n \"llm\": 8008,\n \"image_captioning\": 8007,\n \"exif\": 8010,\n \"tags\": 8011,\n}\n\nHTTP_OK = 200\n\n\ndef check_services():\n for service in SERVICES.keys():\n if service in INCOMPATIBLE_SERVICES:\n logger.info(f\"Skipping restart of incompatible service: {service}\")\n continue\n\n if not is_healthy(service):\n stop_service(service)\n logger.info(f\"Restarting {service}\")\n start_service(service)\n\n\ndef is_healthy(service):\n port = SERVICES.get(service)\n try:\n res = requests.get(f\"http://localhost:{port}/health\")\n # If response has timestamp, check if it needs to be restarted\n if res.json().get(\"last_request_time\") is not None:\n if res.json()[\"last_request_time\"] < time.time() - 120:\n logger.info(f\"Service {service} is stale and needs to be restarted\")\n return False\n return res.status_code == HTTP_OK\n except BaseException as e:\n logger.exception(f\"Error checking health of {service}: {str(e)}\")\n return False\n\n\ndef start_service(service):\n # Check system compatibility before attempting to start the service\n if not is_service_compatible(service):\n logger.error(f\"Service '{service}' is not compatible with this system\")\n return False\n\n if service == \"image_similarity\":\n subprocess.Popen(\n [\n \"python\",\n \"image_similarity/main.py\",\n \"2>&1 | tee /logs/image_similarity.log\",\n ]\n )\n elif service in SERVICES.keys():\n subprocess.Popen(\n [\n \"python\",\n f\"service/{service}/main.py\",\n \"2>&1 | tee /logs/{service}.log\",\n ]\n )\n else:\n logger.warning(\"Unknown service:\", service)\n return False\n\n logger.info(f\"Service '{service}' started successfully\")\n return True\n\n\ndef stop_service(service):\n try:\n # Find the process ID (PID) of the service using `ps` and `grep`\n ps_command = f\"ps aux | grep '[p]ython.*{service}/main.py' | awk '{{print $2}}'\"\n result = subprocess.run(\n ps_command,\n shell=True,\n check=True,\n stdout=subprocess.PIPE,\n stderr=subprocess.PIPE,\n )\n\n pids = result.stdout.decode().strip().split()\n\n if not pids:\n logger.warning(\"Service '%s' is not running\", service)\n return False\n\n # Kill each process found\n for pid in pids:\n subprocess.run([\"kill\", \"-9\", pid], check=True)\n logger.info(f\"Service '{service}' with PID {pid} stopped successfully\")\n\n return True\n except subprocess.CalledProcessError as e:\n logger.error(f\"Failed to stop service '{service}': {e.stderr.decode().strip()}\")\n return False\n except Exception as e:\n logger.error(f\"An error occurred while stopping service '{service}': {e}\")\n return False\n\n\ndef _is_arm_architecture():\n \"\"\"Check if the current system is running on ARM architecture\n \n Returns:\n bool: True if ARM architecture, False otherwise\n \"\"\"\n machine = platform.machine().lower()\n return machine in ['aarch64', 'arm64', 'armv7l', 'armv8']\n\n\ndef check_cpu_features():\n \"\"\"Check for CPU instruction sets for various services\n \n Note: x86/x64-specific instruction sets (AVX, SSE, etc.) only apply to x86/x64 CPUs.\n On ARM architectures, these checks are skipped as they are not relevant.\n \"\"\"\n # Check if we're on ARM architecture\n if _is_arm_architecture():\n machine = platform.machine()\n logger.info(f\"Detected ARM architecture ({machine}), skipping x86-specific CPU feature checks\")\n return [] # Return empty list as x86 features don't apply to ARM\n \n # Features to check for (x86/x64 specific)\n features_to_check = [\"avx\", \"avx2\", \"sse4_2\", \"fma\", \"f16c\"]\n available_features = []\n\n if not available_features:\n try:\n import cpuinfo\n\n cpu_info = cpuinfo.get_cpu_info()\n flags = cpu_info.get(\"flags\", [])\n for feature in features_to_check:\n if feature in flags:\n available_features.append(feature)\n except ImportError:\n pass\n\n return available_features\n\n\ndef has_required_cpu_features(service):\n \"\"\"Check if CPU has required features for a specific service\n \n On ARM architectures, x86-specific CPU checks are bypassed since those\n instruction sets don't exist on ARM. Services like llama.cpp support ARM natively.\n \"\"\"\n if service not in SERVICE_CPU_REQUIREMENTS:\n return True # No CPU requirements for this service\n\n # Check if we're on ARM architecture\n if _is_arm_architecture():\n machine = platform.machine()\n logger.info(f\"Running on ARM architecture ({machine}), skipping x86-specific CPU feature requirements for {service}\")\n return True # Skip x86-specific checks on ARM\n\n requirements = SERVICE_CPU_REQUIREMENTS[service]\n required_features = requirements.get(\"required\", [])\n recommended_features = requirements.get(\"recommended\", [])\n\n available_features = check_cpu_features()\n\n logger.info(f\"CPU features detected for {service}: {available_features}\")\n\n missing_required = []\n missing_recommended = []\n\n for feature in required_features:\n if feature not in available_features:\n missing_required.append(feature)\n\n for feature in recommended_features:\n if feature not in available_features:\n missing_recommended.append(feature)\n\n if missing_required:\n logger.error(f\"Service '{service}' requires CPU features: {missing_required}\")\n logger.error(f\"Missing required CPU features: {missing_required}\")\n return False\n\n if missing_recommended:\n logger.warning(\n f\"Service '{service}' performance may be degraded without: {missing_recommended}\"\n )\n\n logger.info(f\"CPU compatible with service '{service}'\")\n return True\n\n\ndef is_service_compatible(service):\n \"\"\"Check if a service is compatible with the current system\"\"\"\n # Check CPU compatibility\n if not has_required_cpu_features(service):\n INCOMPATIBLE_SERVICES.add(service)\n return False\n\n return True\n\n\ndef cleanup_deleted_photos():\n deleted_photos = Photo.objects.filter(\n Q(removed=True) & Q(last_modified__gte=timezone.now() - timedelta(days=30))\n )\n for photo in deleted_photos:\n photo.delete()\n" }, { "path": "api/social_graph.py", "content": "import networkx as nx\nfrom django.db import connection\n\nfrom api.models import Person\nfrom api.util import logger\n\n\ndef build_social_graph(user):\n try:\n query = \"\"\"\n WITH face AS (\n SELECT photo_id, person_id, name, owner_id\n FROM api_face\n JOIN api_person ON api_person.id = person_id\n JOIN api_photo ON api_photo.id = photo_id\n WHERE person_id IS NOT NULL\n AND owner_id = {}\n )\n SELECT f1.name, f2.name\n FROM face f1\n JOIN face f2 USING (photo_id)\n WHERE f1.person_id != f2.person_id\n GROUP BY f1.name, f2.name\n \"\"\".replace(\"{}\", str(user.id))\n G = nx.Graph()\n with connection.cursor() as cursor:\n cursor.execute(query)\n links = cursor.fetchall()\n if len(links) == 0:\n return {\"nodes\": [], \"links\": []}\n for link in links:\n G.add_edge(link[0], link[1])\n pos = nx.spring_layout(G, k=1 / 2, scale=1000, iterations=20)\n return {\n \"nodes\": [\n {\"id\": node, \"x\": coords[0], \"y\": coords[1]}\n for node, coords in pos.items()\n ],\n \"links\": [{\"source\": pair[0], \"target\": pair[1]} for pair in G.edges()],\n }\n except Exception:\n logger.exception(f\"Error building social graph for user {user.id}\")\n raise\n\n\ndef build_ego_graph(person_id):\n G = nx.Graph()\n person = Person.objects.prefetch_related(\"faces__photo__faces__person\").filter(\n id=person_id\n )[0]\n for this_person_face in person.faces.all():\n for other_person_face in this_person_face.photo.faces.all():\n G.add_edge(person.name, other_person_face.person.name)\n nodes = [{\"id\": node} for node in G.nodes()]\n links = [{\"source\": pair[0], \"target\": pair[1]} for pair in G.edges()]\n res = {\"nodes\": nodes, \"links\": links}\n return res\n" }, { "path": "api/stack_detection.py", "content": "\"\"\"\nStack detection module for grouping related photos organizationally.\n\nHandles organizational stack types:\n- BURST_SEQUENCE: Photos taken in rapid succession\n- EXPOSURE_BRACKET: Bracketed exposures for HDR\n- MANUAL: User-created stacks (not detected, created by user)\n\nNOTE: RAW+JPEG pairs and Live Photos are NO LONGER handled as stacks.\nThey now use the Photo.files ManyToMany field for file variants\n(PhotoPrism-like model). This is handled during scan, not detection.\n\nNOTE: Duplicate detection (exact copies and visual duplicates) is now\nhandled separately by api/duplicate_detection.py. This module focuses\non organizational grouping, not storage cleanup.\n\nBurst detection uses a rules-based system with two categories:\n- Hard criteria: Deterministic (EXIF tags, filename patterns)\n- Soft criteria: Estimation (timestamp proximity, visual similarity)\n\"\"\"\n\nfrom collections import defaultdict\n\nfrom django.db.models import Q\n\nfrom api.models import Photo\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.long_running_job import LongRunningJob\nfrom api.burst_detection_rules import (\n as_rules,\n get_enabled_rules,\n get_hard_rules,\n get_soft_rules,\n group_photos_by_timestamp,\n group_photos_by_visual_similarity,\n BurstRuleTypes,\n)\nfrom api.util import logger\n\n\ndef clear_stacks_of_type(user, stack_type):\n \"\"\"\n Clear all stacks of a specific type for a user before re-detection.\n This ensures we start fresh and don't create duplicate stacks.\n\n Args:\n user: The user whose stacks to clear\n stack_type: The stack type to clear (e.g., PhotoStack.StackType.BURST_SEQUENCE)\n\n Returns:\n Number of stacks deleted\n \"\"\"\n stacks_to_delete = PhotoStack.objects.filter(owner=user, stack_type=stack_type)\n\n count = stacks_to_delete.count()\n\n # Unlink all photos from these stacks (ManyToMany)\n for stack in stacks_to_delete:\n for photo in stack.photos.all():\n photo.stacks.remove(stack)\n\n # Delete the stacks\n stacks_to_delete.delete()\n\n if count > 0:\n logger.info(f\"Cleared {count} {stack_type} stacks for {user.username}\")\n\n return count\n\n\ndef detect_burst_sequences(\n user, interval_ms=2000, use_visual_similarity=True, progress_callback=None\n):\n \"\"\"\n Detect burst sequences using user's configured rules.\n\n This function now uses a rules-based system with two categories:\n - Hard criteria: EXIF tags, filename patterns (deterministic)\n - Soft criteria: Timestamp proximity, visual similarity (estimation)\n\n By default, only hard criteria rules are enabled.\n\n Args:\n user: The user whose photos to analyze\n interval_ms: Default milliseconds between burst photos (for soft rules without config)\n use_visual_similarity: Default for visual similarity (for soft rules without config)\n progress_callback: Optional callback(current, total, found)\n\n Returns:\n Number of stacks created\n \"\"\"\n # Clear existing burst stacks before re-detection\n clear_stacks_of_type(user, PhotoStack.StackType.BURST_SEQUENCE)\n\n # Get user's burst detection rules\n rules_config = user.burst_detection_rules\n if isinstance(rules_config, str):\n import json\n\n rules_config = json.loads(rules_config)\n\n rules = as_rules(rules_config)\n enabled_rules = get_enabled_rules(rules)\n\n if not enabled_rules:\n logger.info(f\"No burst detection rules enabled for {user.username}\")\n return 0\n\n hard_rules = get_hard_rules(rules)\n soft_rules = get_soft_rules(rules)\n\n stacks_created = 0\n\n # === Phase 1: Hard criteria detection ===\n if hard_rules:\n stacks_created += _detect_bursts_hard_criteria(\n user, hard_rules, progress_callback\n )\n\n # === Phase 2: Soft criteria detection ===\n if soft_rules:\n stacks_created += _detect_bursts_soft_criteria(\n user, soft_rules, interval_ms, use_visual_similarity, progress_callback\n )\n\n logger.info(\n f\"Burst detection for {user.username}: found {stacks_created} sequences\"\n )\n return stacks_created\n\n\ndef _detect_bursts_hard_criteria(user, hard_rules, progress_callback=None):\n \"\"\"\n Detect bursts using hard criteria (EXIF tags, filename patterns).\n\n These are deterministic rules that identify burst photos based on\n camera metadata or filename conventions.\n \"\"\"\n from api.metadata.reader import get_metadata\n\n # Get all photos that could be in bursts\n photos = Photo.objects.filter(\n Q(owner=user) & Q(hidden=False) & Q(in_trashcan=False)\n ).select_related(\"main_file\", \"metadata\")\n\n total = photos.count()\n if total == 0:\n return 0\n\n # Collect required EXIF tags from all rules\n required_tags = set()\n for rule in hard_rules:\n required_tags.update(rule.get_required_exif_tags())\n required_tags = list(required_tags)\n\n # Group photos by burst group_key\n burst_groups = defaultdict(list)\n\n for i, photo in enumerate(photos):\n if not photo.main_file:\n continue\n\n # Get EXIF tags for this photo\n try:\n exif_values = get_metadata(photo.main_file.path, required_tags)\n exif_tags = dict(zip(required_tags, exif_values))\n except Exception as e:\n logger.debug(f\"Could not read EXIF for {photo.main_file.path}: {e}\")\n exif_tags = {}\n\n # Try each hard rule until one matches\n for rule in hard_rules:\n is_burst, group_key = rule.is_burst_photo(photo, exif_tags)\n if is_burst and group_key:\n burst_groups[group_key].append(photo)\n break # Photo matched a rule, don't try others\n\n if progress_callback and i % 100 == 0:\n progress_callback(i, total, len(burst_groups))\n\n # Create stacks from groups with 2+ photos\n stacks_created = 0\n for group_key, photos_in_group in burst_groups.items():\n if len(photos_in_group) >= 2:\n # Sort by timestamp if available\n photos_in_group.sort(key=lambda p: p.exif_timestamp or p.added_on)\n stack = _create_burst_stack(user, photos_in_group)\n if stack:\n stacks_created += 1\n logger.debug(\n f\"Created hard-criteria burst stack: {group_key} with {len(photos_in_group)} photos\"\n )\n\n logger.info(\n f\"Hard criteria burst detection: found {stacks_created} stacks from {len(burst_groups)} groups\"\n )\n return stacks_created\n\n\ndef _detect_bursts_soft_criteria(\n user,\n soft_rules,\n default_interval_ms=2000,\n default_use_visual=True,\n progress_callback=None,\n):\n \"\"\"\n Detect bursts using soft criteria (timestamp proximity, visual similarity).\n\n These are estimation-based rules that group photos based on timing\n and/or visual similarity.\n \"\"\"\n # Get photos ordered by timestamp (needed for proximity detection)\n photos = (\n Photo.objects.filter(\n Q(owner=user)\n & Q(exif_timestamp__isnull=False)\n & Q(hidden=False)\n & Q(in_trashcan=False)\n )\n .order_by(\"exif_timestamp\")\n .select_related(\"main_file\", \"metadata\")\n )\n\n total = photos.count()\n if total < 2:\n return 0\n\n stacks_created = 0\n photos_list = list(photos)\n\n # Process each soft rule\n for rule in soft_rules:\n if rule.rule_type == BurstRuleTypes.TIMESTAMP_PROXIMITY:\n # Get rule-specific parameters or use defaults\n interval_ms = rule.params.get(\"interval_ms\", default_interval_ms)\n require_same_camera = rule.params.get(\"require_same_camera\", True)\n\n groups = group_photos_by_timestamp(\n photos_list, interval_ms, require_same_camera\n )\n\n for group in groups:\n # Filter out photos already in burst stacks\n photos_to_stack = [\n p\n for p in group\n if not p.stacks.filter(\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n ).exists()\n ]\n if len(photos_to_stack) >= 2:\n stack = _create_burst_stack(user, photos_to_stack)\n if stack:\n stacks_created += 1\n\n elif rule.rule_type == BurstRuleTypes.VISUAL_SIMILARITY:\n similarity_threshold = rule.params.get(\"similarity_threshold\", 15)\n\n groups = group_photos_by_visual_similarity(\n photos_list, similarity_threshold\n )\n\n for group in groups:\n # Filter out photos already in burst stacks\n photos_to_stack = [\n p\n for p in group\n if not p.stacks.filter(\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n ).exists()\n ]\n if len(photos_to_stack) >= 2:\n stack = _create_burst_stack(user, photos_to_stack)\n if stack:\n stacks_created += 1\n\n logger.info(f\"Soft criteria burst detection: found {stacks_created} stacks\")\n return stacks_created\n\n\ndef _create_burst_stack(user, photos):\n \"\"\"Helper to create a burst stack from a list of photos.\"\"\"\n if len(photos) < 2:\n return None\n\n # Filter out photos that are already in a burst stack to prevent duplicates\n # A photo should only be in one burst stack at a time\n photos_to_stack = [\n photo\n for photo in photos\n if not photo.stacks.filter(\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n ).exists()\n ]\n\n # If all photos are already stacked, skip\n if len(photos_to_stack) < 2:\n return None\n\n # Use create_or_merge to ensure photos aren't in multiple stacks of the same type\n # Pass sequence timestamps for burst stacks\n stack = PhotoStack.create_or_merge(\n owner=user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n photos=photos_to_stack,\n sequence_start=photos_to_stack[0].exif_timestamp,\n sequence_end=photos_to_stack[-1].exif_timestamp,\n )\n\n logger.info(\n f\"Created/merged BURST_SEQUENCE stack with {len(photos_to_stack)} photos\"\n )\n return stack\n\n\ndef batch_detect_stacks(user, options=None):\n \"\"\"\n Run batch stack detection for a user.\n\n NOTE: RAW+JPEG pairs and Live Photos are now handled as file variants\n during scan (PhotoPrism-like model), not as stacks here.\n\n Burst detection uses the user's configured burst_detection_rules from their profile.\n\n Args:\n user: The user whose photos to analyze\n options: Dict with detection options:\n - detect_bursts: bool (default: True) - uses user's burst_detection_rules\n \"\"\"\n if options is None:\n options = {}\n\n detect_bursts = options.get(\"detect_bursts\", True)\n\n # Create long-running job for progress tracking\n job = LongRunningJob.create_job(\n user=user,\n job_type=LongRunningJob.JOB_SCAN_PHOTOS,\n start_now=True,\n )\n\n try:\n total_found = 0\n\n # Detect burst sequences (uses user's burst_detection_rules)\n if detect_bursts:\n\n def progress_burst(current, total, found):\n job.set_result(\n {\n \"stage\": \"burst_sequences\",\n \"current\": current,\n \"total\": total,\n \"found\": found,\n }\n )\n\n burst_count = detect_burst_sequences(user, progress_callback=progress_burst)\n total_found += burst_count\n\n job.complete(result={\"status\": \"completed\", \"stacks_found\": total_found})\n\n logger.info(\n f\"Stack detection completed for {user.username}: {total_found} stacks found\"\n )\n\n except Exception as e:\n logger.error(f\"Stack detection failed for {user.username}: {e}\")\n job.fail(error=e)\n raise\n" }, { "path": "api/stacks/__init__.py", "content": "\"\"\"\nPhoto stacking detection and management.\n\nThis package provides unified photo grouping functionality:\n- Exact copies (same MD5 hash)\n- Visual duplicates (similar pHash/CLIP)\n- RAW+JPEG pairs\n- Burst sequences\n- Exposure brackets\n- Live Photos (embedded motion video)\n- Manual user groupings\n\"\"\"\n\nfrom api.stacks.live_photo import (\n detect_live_photo,\n extract_embedded_motion_video,\n find_apple_live_photo_video,\n has_embedded_motion_video,\n process_live_photos_batch,\n)\n\n__all__ = [\n \"detect_live_photo\",\n \"extract_embedded_motion_video\",\n \"find_apple_live_photo_video\",\n \"has_embedded_motion_video\",\n \"process_live_photos_batch\",\n]\n" }, { "path": "api/stacks/live_photo.py", "content": "\"\"\"\nLive Photo detection and stacking logic.\n\nHandles extraction and grouping of Live Photos:\n- Google Pixel Motion Photos (embedded MP4 after JPEG EOI)\n- Samsung Motion Photos (MotionPhoto_Data marker)\n- Apple Live Photos (paired .mov file)\n\nThis module moves embedded media extraction from directory_watcher to \na dedicated stacks-aware component for better organization.\n\"\"\"\n\nfrom mmap import ACCESS_READ, mmap\nfrom pathlib import Path\nfrom typing import TYPE_CHECKING\n\nimport magic\nfrom django.conf import settings\n\nfrom api.models.file import File\nfrom api.models.photo_stack import PhotoStack\nfrom api.util import logger\n\nif TYPE_CHECKING:\n from api.models.photo import Photo\n from api.models.user import User\n\n\n# Markers for embedded motion video detection\nJPEG_EOI_MARKER = b\"\\xff\\xd9\"\nGOOGLE_PIXEL_MP4_SIGNATURES = [b\"ftypmp42\", b\"ftypisom\", b\"ftypiso2\"]\nSAMSUNG_MOTION_MARKER = b\"MotionPhoto_Data\"\n\n# Apple Live Photo video extensions\nAPPLE_LIVE_PHOTO_EXTENSIONS = [\".mov\", \".MOV\"]\n\n\ndef _locate_google_embedded_video(data: bytes) -> int:\n \"\"\"Find position of embedded MP4 in Google Motion Photo.\"\"\"\n for signature in GOOGLE_PIXEL_MP4_SIGNATURES:\n position = data.find(signature)\n if position != -1:\n # MP4 header starts 4 bytes before ftyp\n return position - 4\n return -1\n\n\ndef _locate_samsung_embedded_video(data: bytes) -> int:\n \"\"\"Find position of embedded video in Samsung Motion Photo.\"\"\"\n position = data.find(SAMSUNG_MOTION_MARKER)\n if position != -1:\n # Video starts immediately after the marker\n return position + len(SAMSUNG_MOTION_MARKER)\n return -1\n\n\ndef has_embedded_motion_video(path: str) -> bool:\n \"\"\"\n Check if a JPEG file contains an embedded motion video.\n \n Supports:\n - Google Pixel Motion Photos\n - Samsung Motion Photos\n \n Args:\n path: Path to the image file\n \n Returns:\n True if embedded video detected, False otherwise\n \"\"\"\n try:\n mime = magic.Magic(mime=True)\n mime_type = mime.from_file(path)\n if mime_type != \"image/jpeg\":\n return False\n \n with open(path, \"rb\") as image:\n with mmap(image.fileno(), 0, access=ACCESS_READ) as mm:\n return (\n _locate_google_embedded_video(mm) != -1 or\n _locate_samsung_embedded_video(mm) != -1\n )\n except Exception as e:\n logger.warning(f\"Error checking for embedded video in {path}: {e}\")\n return False\n\n\ndef extract_embedded_motion_video(path: str, output_hash: str) -> str | None:\n \"\"\"\n Extract embedded motion video from a JPEG file.\n \n Args:\n path: Path to the source image file\n output_hash: Hash to use for output filename\n \n Returns:\n Path to extracted video file, or None if extraction failed\n \"\"\"\n try:\n with open(str(path), \"rb\") as image:\n with mmap(image.fileno(), 0, access=ACCESS_READ) as mm:\n # Try Google format first, then Samsung\n position = _locate_google_embedded_video(mm)\n if position == -1:\n position = _locate_samsung_embedded_video(mm)\n \n if position == -1:\n return None\n \n # Create output directory\n output_dir = Path(settings.MEDIA_ROOT) / \"embedded_media\"\n output_dir.mkdir(parents=True, exist_ok=True)\n \n output_path = output_dir / f\"{output_hash}_motion.mp4\"\n \n with open(output_path, \"wb\") as video:\n mm.seek(position)\n data = mm.read(mm.size() - position)\n video.write(data)\n \n logger.info(f\"Extracted motion video to {output_path}\")\n return str(output_path)\n \n except Exception as e:\n logger.error(f\"Error extracting embedded video from {path}: {e}\")\n return None\n\n\ndef find_apple_live_photo_video(image_path: str) -> str | None:\n \"\"\"\n Find the companion .mov file for an Apple Live Photo.\n \n Apple Live Photos are stored as separate .HEIC/.JPG and .MOV files\n with the same base name (from ContentIdentifier).\n \n Args:\n image_path: Path to the image file\n \n Returns:\n Path to companion video file, or None if not found\n \"\"\"\n base_path = Path(image_path)\n stem = base_path.stem\n parent = base_path.parent\n \n for ext in APPLE_LIVE_PHOTO_EXTENSIONS:\n video_path = parent / f\"{stem}{ext}\"\n if video_path.exists():\n return str(video_path)\n \n return None\n\n\ndef detect_live_photo(photo: \"Photo\", user: \"User\") -> PhotoStack | None:\n \"\"\"\n Detect if a photo is part of a Live Photo and create a stack.\n \n This handles:\n 1. Embedded motion videos (Google/Samsung) - extracts and links\n 2. Apple Live Photos - finds and links companion video\n \n Args:\n photo: The Photo instance to check\n user: Owner of the photo\n \n Returns:\n PhotoStack instance if Live Photo detected, None otherwise\n \"\"\"\n if not photo.main_file:\n return None\n \n image_path = photo.main_file.path\n \n # Check for embedded motion video (Google/Samsung)\n if has_embedded_motion_video(image_path):\n return _create_embedded_live_photo_stack(photo, user)\n \n # Check for Apple Live Photo companion video\n video_path = find_apple_live_photo_video(image_path)\n if video_path:\n return _create_apple_live_photo_stack(photo, video_path, user)\n \n return None\n\n\ndef _create_embedded_live_photo_stack(photo: \"Photo\", user: \"User\") -> PhotoStack | None:\n \"\"\"Create stack for photo with embedded motion video.\"\"\"\n if not settings.FEATURE_PROCESS_EMBEDDED_MEDIA:\n logger.debug(\"Embedded media processing disabled\")\n return None\n \n image_path = photo.main_file.path\n video_path = extract_embedded_motion_video(image_path, photo.main_file.hash)\n \n if not video_path:\n return None\n \n # Create File record for the extracted video\n video_file = File.create(video_path, user)\n \n # Link as embedded media on the original file\n photo.main_file.embedded_media.add(video_file)\n \n # Create or update stack\n existing_stack = photo.stacks.filter(stack_type=PhotoStack.StackType.LIVE_PHOTO).first()\n if existing_stack:\n return existing_stack\n \n # Create new Live Photo stack\n stack = PhotoStack.objects.create(\n owner=user,\n stack_type=PhotoStack.StackType.LIVE_PHOTO,\n primary_photo=photo,\n )\n \n # Link photo to stack (ManyToMany)\n photo.stacks.add(stack)\n \n logger.info(f\"Created Live Photo stack {stack.id} for embedded motion in {image_path}\")\n return stack\n\n\ndef _create_apple_live_photo_stack(\n photo: \"Photo\", \n video_path: str, \n user: \"User\"\n) -> PhotoStack | None:\n \"\"\"Create stack for Apple Live Photo with companion video.\"\"\"\n from api.models.photo import Photo\n \n # Check if video is already a known photo/file\n video_file = File.objects.filter(path=video_path).first()\n \n if not video_file:\n # Create File record for the video\n video_file = File.create(video_path, user)\n \n # Find or create the video as a Photo\n video_photo = Photo.objects.filter(main_file=video_file).first()\n \n if not video_photo:\n # Video file exists but no Photo record - might be created by scan\n # Link as embedded media instead\n photo.main_file.embedded_media.add(video_file)\n \n # Create or find stack\n existing_stack = photo.stacks.filter(stack_type=PhotoStack.StackType.LIVE_PHOTO).first()\n if existing_stack:\n stack = existing_stack\n else:\n stack = PhotoStack.objects.create(\n owner=user,\n stack_type=PhotoStack.StackType.LIVE_PHOTO,\n primary_photo=photo,\n )\n photo.stacks.add(stack)\n \n # If video is a separate Photo, link it to the same stack\n if video_photo and not video_photo.stacks.filter(stack_type=PhotoStack.StackType.LIVE_PHOTO).exists():\n video_photo.stacks.add(stack)\n \n logger.info(f\"Created Apple Live Photo stack {stack.id} for {photo.main_file.path}\")\n return stack\n\n\ndef process_live_photos_batch(user: \"User\", photos: list[\"Photo\"]) -> dict:\n \"\"\"\n Process multiple photos for Live Photo detection.\n \n Args:\n user: User who owns the photos\n photos: List of Photo instances to check\n \n Returns:\n Dict with counts: {detected: int, stacks_created: int}\n \"\"\"\n detected = 0\n stacks_created = 0\n \n for photo in photos:\n try:\n stack = detect_live_photo(photo, user)\n if stack:\n detected += 1\n if stack.photo_count <= 1:\n # New stack (might just have the photo, video linked separately)\n stacks_created += 1\n except Exception as e:\n logger.error(f\"Error processing Live Photo detection for {photo.id}: {e}\")\n \n return {\n \"detected\": detected,\n \"stacks_created\": stacks_created,\n }\n" }, { "path": "api/stats.py", "content": "import os\nfrom datetime import datetime\n\nimport numpy as np\nfrom django.db import connection\nfrom django.db.models import Avg, Count, Max, Min, Q, Sum\nfrom django.db.models.functions import TruncMonth\n\nimport random\nimport re\n\nimport seaborn as sns\nfrom api.util import logger\n\nfrom api.models import (\n AlbumAuto,\n AlbumDate,\n AlbumPlace,\n AlbumThing,\n AlbumUser,\n Cluster,\n Face,\n Person,\n Photo,\n User,\n)\nfrom api.models.user import get_deleted_user\n\n\ndef _is_sqlite() -> bool:\n return connection.vendor == \"sqlite\"\n\n\ndef jump_by_month(start_date, end_date, month_step=1):\n current_date = start_date\n yield current_date\n while current_date < end_date:\n carry, new_month = divmod(current_date.month - 1 + month_step, 12)\n new_month += 1\n current_date = current_date.replace(\n year=current_date.year + carry, month=new_month\n )\n yield current_date\n\n\ndef median_value(queryset, term):\n from decimal import Decimal\n\n count = queryset.count()\n if count == 0:\n return\n values = queryset.values_list(term, flat=True).order_by(term)\n if count % 2 == 1:\n return values[int(round(count / 2))]\n else:\n return sum(values[count / 2 - 1 : count / 2 + 1]) / Decimal(2.0)\n\n\ndef calc_megabytes(bytes):\n if bytes == 0 or bytes is None:\n return 0\n return round((bytes / 1024) / 1024)\n\n\ndef get_server_stats():\n # CPU architecture, Speed, Number of Cores, 64bit / 32 Bits\n import cpuinfo\n\n cpu_info = cpuinfo.get_cpu_info()\n # Available RAM\n import psutil\n\n available_ram = calc_megabytes(psutil.virtual_memory().total)\n # GPU\n import torch\n\n if torch.cuda.is_available():\n gpu_name = torch.cuda.get_device_name(0)\n gpu_memory = calc_megabytes(torch.cuda.get_device_properties(0).total_memory)\n else:\n gpu_name = \"\"\n gpu_memory = \"\"\n # Total Capacity\n import shutil\n\n total_storage, used_storage, free_storage = shutil.disk_usage(\"/\")\n image_tag = os.environ.get(\"IMAGE_TAG\", \"\")\n number_of_users = User.objects.filter(~Q(id=get_deleted_user().id)).count()\n users = []\n for user in User.objects.filter(~Q(id=get_deleted_user().id)):\n date_joined = user.date_joined\n number_of_photos = Photo.objects.filter(Q(owner=user)).count()\n number_of_videos = Photo.objects.filter(Q(owner=user) & Q(video=True)).count()\n number_of_captions = Photo.objects.filter(\n Q(owner=user)\n & Q(caption_instance__captions_json__user_caption__isnull=False)\n ).count()\n number_of_generated_captions = Photo.objects.filter(\n Q(owner=user) & Q(caption_instance__captions_json__im2txt__isnull=False)\n ).count()\n number_of_albums = AlbumUser.objects.filter(Q(owner=user)).count()\n min_number_of_photos_per_album = (\n AlbumUser.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Min(\"count\"))\n )\n max_number_of_photos_per_album = (\n AlbumUser.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_photos_per_album = (\n AlbumUser.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_photos_per_album = median_value(\n AlbumUser.objects.filter(Q(owner=user)).annotate(count=Count(\"photos\")),\n \"count\",\n )\n min_number_of_videos_per_album = (\n AlbumUser.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Min(\"count\"))\n )\n max_number_of_videos_per_album = (\n AlbumUser.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_videos_per_album = (\n AlbumUser.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_videos_per_album = median_value(\n AlbumUser.objects.filter(Q(owner=user)).annotate(\n count=Count(\"photos\", filter=Q(photos__video=True))\n ),\n \"count\",\n )\n number_of_persons = Person.objects.filter(Q(cluster_owner=user)).count()\n min_number_of_faces_per_person = (\n Person.objects.filter(Q(cluster_owner=user))\n .annotate(count=Count(\"faces\"))\n .aggregate(Min(\"count\"))\n )\n max_number_of_faces_per_person = (\n Person.objects.filter(Q(cluster_owner=user))\n .annotate(count=Count(\"faces\"))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_faces_per_person = (\n Person.objects.filter(Q(cluster_owner=user))\n .annotate(count=Count(\"faces\"))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_faces_per_person = median_value(\n Person.objects.filter(Q(cluster_owner=user)).annotate(count=Count(\"faces\")),\n \"count\",\n )\n number_of_clusters = Cluster.objects.filter(Q(owner=user)).count()\n number_of_places = AlbumPlace.objects.filter(Q(owner=user)).count()\n min_number_of_photos_per_place = (\n AlbumPlace.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Min(\"count\"))\n )\n max_number_of_photos_per_place = (\n AlbumPlace.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_photos_per_place = (\n AlbumPlace.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_photos_per_place = median_value(\n AlbumPlace.objects.filter(Q(owner=user)).annotate(count=Count(\"photos\")),\n \"count\",\n )\n min_number_of_videos_per_place = (\n AlbumPlace.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Min(\"count\"))\n )\n max_number_of_videos_per_place = (\n AlbumPlace.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_videos_per_place = (\n AlbumPlace.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_videos_per_place = median_value(\n AlbumPlace.objects.filter(Q(owner=user)).annotate(\n count=Count(\"photos\", filter=Q(photos__video=True))\n ),\n \"count\",\n )\n number_of_things = AlbumThing.objects.filter(Q(owner=user)).count()\n min_number_of_photos_per_thing = (\n AlbumThing.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Min(\"count\"))\n )\n max_number_of_photos_per_thing = (\n AlbumThing.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_photos_per_thing = (\n AlbumThing.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_photos_per_thing = median_value(\n AlbumThing.objects.filter(Q(owner=user)).annotate(count=Count(\"photos\")),\n \"count\",\n )\n min_number_of_videos_per_thing = (\n AlbumThing.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Min(\"count\"))\n )\n max_number_of_videos_per_thing = (\n AlbumThing.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_videos_per_thing = (\n AlbumThing.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_videos_per_thing = median_value(\n AlbumThing.objects.filter(Q(owner=user)).annotate(\n count=Count(\"photos\", filter=Q(photos__video=True))\n ),\n \"count\",\n )\n number_of_events = AlbumAuto.objects.filter(Q(owner=user)).count()\n min_number_of_photos_per_event = (\n AlbumAuto.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Min(\"count\"))\n )\n max_number_of_photos_per_event = (\n AlbumAuto.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_photos_per_event = (\n AlbumAuto.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\"))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_photos_per_event = median_value(\n AlbumAuto.objects.filter(Q(owner=user)).annotate(count=Count(\"photos\")),\n \"count\",\n )\n min_number_of_videos_per_event = (\n AlbumAuto.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Min(\"count\"))\n )\n max_number_of_videos_per_event = (\n AlbumAuto.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Max(\"count\"))\n )\n mean_number_of_videos_per_event = (\n AlbumAuto.objects.filter(Q(owner=user))\n .annotate(count=Count(\"photos\", filter=Q(photos__video=True)))\n .aggregate(Avg(\"count\"))\n )\n median_number_of_videos_per_event = median_value(\n AlbumAuto.objects.filter(Q(owner=user)).annotate(\n count=Count(\"photos\", filter=Q(photos__video=True))\n ),\n \"count\",\n )\n number_of_favorites = Photo.objects.filter(\n Q(owner=user) & Q(rating__gte=user.favorite_min_rating)\n ).count()\n number_of_hidden = Photo.objects.filter(Q(owner=user) & Q(hidden=True)).count()\n number_of_public = Photo.objects.filter(Q(owner=user) & Q(public=True)).count()\n users.append(\n {\n \"date_joined\": date_joined.strftime(\"%d-%m-%Y\"),\n \"total_file_size_in_mb\": calc_megabytes(\n Photo.objects.filter(Q(owner=user)).aggregate(Sum(\"size\"))[\n \"size__sum\"\n ]\n or None\n ),\n \"number_of_photos\": number_of_photos,\n \"number_of_videos\": number_of_videos,\n \"number_of_captions\": number_of_captions,\n \"number_of_generated_captions\": number_of_generated_captions,\n \"album\": {\n \"count\": number_of_albums,\n \"min\": min_number_of_photos_per_album[\"count__min\"] or None,\n \"max\": max_number_of_photos_per_album[\"count__max\"] or None,\n \"mean\": mean_number_of_photos_per_album[\"count__avg\"] or None,\n \"median\": median_number_of_photos_per_album,\n \"min_videos\": min_number_of_videos_per_album[\"count__min\"] or None,\n \"max_videos\": max_number_of_videos_per_album[\"count__max\"] or None,\n \"mean_videos\": mean_number_of_videos_per_album[\"count__avg\"]\n or None,\n \"median_videos\": median_number_of_videos_per_album,\n },\n \"person\": {\n \"count\": number_of_persons,\n \"min\": min_number_of_faces_per_person[\"count__min\"] or None,\n \"max\": max_number_of_faces_per_person[\"count__max\"] or None,\n \"mean\": mean_number_of_faces_per_person[\"count__avg\"] or None,\n \"median\": median_number_of_faces_per_person,\n },\n \"number_of_clusters\": number_of_clusters,\n \"places\": {\n \"count\": number_of_places,\n \"min\": min_number_of_photos_per_place[\"count__min\"] or None,\n \"max\": max_number_of_photos_per_place[\"count__max\"] or None,\n \"mean\": mean_number_of_photos_per_place[\"count__avg\"] or None,\n \"median\": median_number_of_photos_per_place,\n \"min_videos\": min_number_of_videos_per_place[\"count__min\"] or None,\n \"max_videos\": max_number_of_videos_per_place[\"count__max\"] or None,\n \"mean_videos\": mean_number_of_videos_per_place[\"count__avg\"]\n or None,\n \"median_videos\": median_number_of_videos_per_place,\n },\n \"things\": {\n \"count\": number_of_things,\n \"min\": min_number_of_photos_per_thing[\"count__min\"] or None,\n \"max\": max_number_of_photos_per_thing[\"count__max\"] or None,\n \"mean\": mean_number_of_photos_per_thing[\"count__avg\"] or None,\n \"median\": median_number_of_photos_per_thing,\n \"min_videos\": min_number_of_videos_per_thing[\"count__min\"] or None,\n \"max_videos\": max_number_of_videos_per_thing[\"count__max\"] or None,\n \"mean_videos\": mean_number_of_videos_per_thing[\"count__avg\"]\n or None,\n \"median_videos\": median_number_of_videos_per_thing,\n },\n \"events\": {\n \"count\": number_of_events,\n \"min\": min_number_of_photos_per_event[\"count__min\"] or None,\n \"max\": max_number_of_photos_per_event[\"count__max\"] or None,\n \"mean\": mean_number_of_photos_per_event[\"count__avg\"] or None,\n \"median\": median_number_of_photos_per_event,\n \"min_videos\": min_number_of_videos_per_event[\"count__min\"] or None,\n \"max_videos\": max_number_of_videos_per_event[\"count__max\"] or None,\n \"mean_videos\": mean_number_of_videos_per_event[\"count__avg\"]\n or None,\n \"median_videos\": median_number_of_videos_per_event,\n },\n \"number_of_favorites\": number_of_favorites,\n \"number_of_hidden\": number_of_hidden,\n \"number_of_public\": number_of_public,\n }\n )\n res = {\n \"cpu_info\": cpu_info,\n \"image_tag\": image_tag,\n \"available_ram_in_mb\": available_ram,\n \"gpu_name\": gpu_name,\n \"gpu_memory_in_mb\": gpu_memory,\n \"total_storage_in_mb\": calc_megabytes(total_storage),\n \"used_storage_in_mb\": calc_megabytes(used_storage),\n \"free_storage_in_mb\": calc_megabytes(free_storage),\n \"number_of_users\": number_of_users,\n \"users\": users,\n }\n return res\n\n\ndef get_count_stats(user):\n num_photos = Photo.visible.filter(Q(owner=user)).distinct().count()\n num_missing_photos = Photo.objects.filter(\n Q(owner=user) & Q(files=None) | Q(main_file=None)\n ).count()\n num_faces = Face.objects.filter(photo__owner=user).count()\n num_unknown_faces = Face.objects.filter(\n (\n Q(person__name__exact=\"unknown\")\n | Q(person__name__exact=Person.UNKNOWN_PERSON_NAME)\n )\n & Q(photo__owner=user)\n ).count()\n num_labeled_faces = Face.objects.filter(\n Q(person__isnull=False) & Q(photo__owner=user) & Q(photo__hidden=False)\n ).count()\n num_inferred_faces = Face.objects.filter(\n Q(person=True) & Q(photo__owner=user) & Q(photo__hidden=False)\n ).count()\n num_people = (\n Person.objects.filter(\n Q(faces__photo__hidden=False)\n & Q(faces__photo__owner=user)\n & Q(faces__person__isnull=False)\n )\n .distinct()\n .annotate(viewable_face_count=Count(\"faces\"))\n .filter(Q(viewable_face_count__gt=0))\n .count()\n )\n num_albumauto = (\n AlbumAuto.objects.filter(owner=user)\n .annotate(photo_count=Count(\"photos\"))\n .filter(Q(photo_count__gt=0))\n .count()\n )\n num_albumdate = (\n AlbumDate.objects.filter(owner=user)\n .annotate(photo_count=Count(\"photos\"))\n .filter(Q(photo_count__gt=0))\n .count()\n )\n num_albumuser = (\n AlbumUser.objects.filter(owner=user)\n .annotate(photo_count=Count(\"photos\"))\n .filter(Q(photo_count__gt=0))\n .count()\n )\n\n res = {\n \"num_photos\": num_photos,\n \"num_missing_photos\": num_missing_photos,\n \"num_faces\": num_faces,\n \"num_people\": num_people,\n \"num_unknown_faces\": num_unknown_faces,\n \"num_labeled_faces\": num_labeled_faces,\n \"num_inferred_faces\": num_inferred_faces,\n \"num_albumauto\": num_albumauto,\n \"num_albumdate\": num_albumdate,\n \"num_albumuser\": num_albumuser,\n }\n return res\n\n\ndef get_photo_month_counts(user):\n counts = (\n Photo.objects.filter(owner=user)\n .exclude(exif_timestamp=None)\n .annotate(month=TruncMonth(\"exif_timestamp\"))\n .values(\"month\")\n .annotate(c=Count(\"image_hash\"))\n .values(\"month\", \"c\")\n )\n\n all_months = [\n c[\"month\"]\n for c in counts\n if c[\"month\"].year >= 2000 and c[\"month\"].year <= datetime.now().year\n ]\n\n if len(all_months) > 0:\n first_month = min(all_months)\n last_month = max(all_months)\n\n month_span = jump_by_month(first_month, last_month)\n counts = sorted(counts, key=lambda k: k[\"month\"])\n\n res = []\n for count in counts:\n key = \"-\".join([str(count[\"month\"].year), str(count[\"month\"].month)])\n count = count[\"c\"]\n res.append([key, count])\n res = dict(res)\n\n out = []\n for month in month_span:\n m = \"-\".join([str(month.year), str(month.month)])\n if m in res.keys():\n out.append({\"month\": m, \"count\": res[m]})\n else:\n out.append({\"month\": m, \"count\": 0})\n\n return out\n else:\n return []\n\n\ndef get_searchterms_wordcloud(user):\n # Python fallbacks (SQLite): stream and aggregate\n from collections import Counter\n\n out = {\"captions\": [], \"people\": [], \"locations\": []}\n\n # Captions: use Places365 categories, attributes and environment from captions_json\n captions_counter: Counter[str] = Counter()\n captions_first_seen: dict[str, int] = {}\n order_index = 0\n captions_iter = (\n Photo.objects.filter(owner=user)\n .exclude(caption_instance__captions_json__isnull=True)\n .values_list(\"caption_instance__captions_json\", flat=True)\n .iterator(chunk_size=2000)\n )\n for caps in captions_iter:\n try:\n places365 = (caps or {}).get(\"places365\", {})\n categories = places365.get(\"categories\", [])\n if isinstance(categories, list):\n for cat in categories:\n if not cat:\n continue\n label = str(cat)\n captions_counter[label] += 1\n if label not in captions_first_seen:\n captions_first_seen[label] = order_index\n order_index += 1\n attributes = places365.get(\"attributes\", [])\n if isinstance(attributes, list):\n for attr in attributes:\n if not attr:\n continue\n label = str(attr)\n captions_counter[label] += 1\n if label not in captions_first_seen:\n captions_first_seen[label] = order_index\n order_index += 1\n environment = places365.get(\"environment\")\n if isinstance(environment, str) and environment:\n label = environment\n captions_counter[label] += 1\n if label not in captions_first_seen:\n captions_first_seen[label] = order_index\n order_index += 1\n except Exception:\n continue\n\n # People: aggregate with ORM to avoid per-row Python loops\n people_rows = (\n Face.objects.filter(photo__owner=user, person__name__isnull=False)\n .values(\"person__name\")\n .annotate(c=Count(\"id\"))\n .order_by(\"-c\")[:100]\n )\n\n # Locations: parse geolocation_json, ignore postcode and poi, one word per photo\n locations_counter: Counter[str] = Counter()\n locations_first_seen: dict[str, int] = {}\n geo_iter = (\n Photo.objects.filter(owner=user)\n .exclude(geolocation_json=None)\n .values_list(\"image_hash\", \"geolocation_json\")\n .iterator(chunk_size=2000)\n )\n for _image_hash, geo in geo_iter:\n try:\n features = (geo or {}).get(\"features\", [])\n except Exception:\n features = []\n seen_values = set()\n for feature in features:\n if not isinstance(feature, dict):\n continue\n place_type = feature.get(\"place_type\")\n value = feature.get(\"text\")\n if not value:\n continue\n # place_type can be list or string\n types = place_type if isinstance(place_type, list) else [place_type]\n types = [t for t in types if t]\n if any(t in (\"postcode\", \"poi\") for t in types):\n continue\n seen_values.add(str(value))\n for value in seen_values:\n locations_counter[value] += 1\n if value not in locations_first_seen:\n locations_first_seen[value] = captions_first_seen.get(\n value, order_index\n )\n order_index += 1\n\n # Build outputs (log of count as before)\n # Ensure stable order to match expectations: attributes/environment may come first\n # Sort captions by count desc, then by first-seen order for deterministic ties\n captions_sorted = sorted(\n captions_counter.items(),\n key=lambda kv: (-kv[1], captions_first_seen.get(kv[0], 1_000_000)),\n )[:100]\n for label, count in captions_sorted:\n out[\"captions\"].append({\"label\": label, \"y\": float(np.log(count))})\n for row in people_rows:\n out[\"people\"].append(\n {\"label\": row[\"person__name\"], \"y\": float(np.log(row[\"c\"]))}\n )\n locations_sorted = sorted(\n locations_counter.items(),\n key=lambda kv: (-kv[1], locations_first_seen.get(kv[0], 1_000_000)),\n )[:100]\n for label, count in locations_sorted:\n out[\"locations\"].append({\"label\": label, \"y\": float(np.log(count))})\n\n return out\n\n\ndef get_location_sunburst(user):\n levels = []\n\n from collections import Counter\n\n counter = Counter()\n # Stream results to avoid caching entire queryset in memory\n photo_geo_iter = (\n Photo.objects.filter(owner=user)\n .exclude(geolocation_json=None)\n .values_list(\"geolocation_json\", flat=True)\n .iterator(chunk_size=2000)\n )\n for geo in photo_geo_iter:\n try:\n features = (geo or {}).get(\"features\", [])\n except Exception:\n features = []\n if not isinstance(features, list) or len(features) < 3:\n continue\n f1 = features[-1] if isinstance(features[-1], dict) else {}\n f2 = features[-2] if isinstance(features[-2], dict) else {}\n f3 = features[-3] if isinstance(features[-3], dict) else {}\n l1 = f1.get(\"text\")\n l2 = f2.get(\"text\")\n l3 = f3.get(\"text\")\n if l1 is None or l2 is None or l3 is None:\n continue\n counter[(l1, l2, l3)] += 1\n levels = [[k[0], k[1], k[2], v] for k, v in counter.items()]\n levels = sorted(levels, key=lambda x: (x[0], x[1], x[2]))\n\n data_structure = {\"name\": \"Places I've visited\", \"children\": []}\n palette = sns.color_palette(\"hls\", 10).as_hex()\n\n for data in levels:\n depth_cursor = data_structure[\"children\"]\n for i, item in enumerate(data[0:-2]):\n idx = None\n j = None\n for j, c in enumerate(depth_cursor):\n if item in c.values():\n idx = j\n if idx is None:\n depth_cursor.append(\n {\"name\": item, \"children\": [], \"hex\": random.choice(palette)}\n )\n idx = len(depth_cursor) - 1\n\n depth_cursor = depth_cursor[idx][\"children\"]\n if i == len(data) - 3:\n depth_cursor.append(\n {\n \"name\": data[-2],\n \"value\": data[-1],\n \"hex\": random.choice(palette),\n }\n )\n\n return data_structure\n\n\ndef get_location_clusters(user):\n start = datetime.now()\n # Build clusters in Python from JSON fields (works for both SQLite and Postgres)\n results_by_location = {}\n # Stream results to avoid large memory usage\n photo_geo_iter = (\n Photo.objects.filter(owner=user)\n .exclude(geolocation_json=None)\n .values_list(\"geolocation_json\", flat=True)\n .iterator(chunk_size=2000)\n )\n numeric_pattern = re.compile(r\"^(-)?[0-9]+$\")\n for geo in photo_geo_iter:\n try:\n features = (geo or {}).get(\"features\", [])\n except Exception:\n features = []\n for feature in features:\n location_text = feature.get(\"text\") if isinstance(feature, dict) else None\n if not location_text or numeric_pattern.match(str(location_text)):\n continue\n center = feature.get(\"center\") if isinstance(feature, dict) else None\n if not (isinstance(center, (list, tuple)) and len(center) >= 2):\n continue\n # Keep first occurrence per distinct location name\n if location_text not in results_by_location:\n lon = center[0]\n lat = center[1]\n try:\n lat_f = float(lat)\n lon_f = float(lon)\n except Exception:\n continue\n results_by_location[location_text] = [lat_f, lon_f, location_text]\n\n # Order by location to mimic SQL ordering\n res = [results_by_location[key] for key in sorted(results_by_location.keys())]\n elapsed = (datetime.now() - start).total_seconds()\n logger.info(\"location clustering computed in %.2f seconds\" % elapsed)\n return res\n\n\ndef get_location_timeline(user):\n # Python fallback: iterate photos ordered by timestamp and build contiguous location spans\n def extract_location(geo: dict) -> str | None:\n if not geo or not isinstance(geo, dict):\n return None\n features = geo.get(\"features\", [])\n if not isinstance(features, list) or not features:\n return None\n last = features[-1]\n if isinstance(last, dict):\n return last.get(\"text\")\n return None\n\n # Stream through photos ordered by exif_timestamp\n qs = (\n Photo.objects.filter(owner=user)\n .exclude(exif_timestamp=None)\n .order_by(\"exif_timestamp\")\n .values_list(\"geolocation_json\", \"exif_timestamp\")\n .iterator(chunk_size=2000)\n )\n spans: list[tuple[str, datetime, datetime]] = []\n current_loc: str | None = None\n run_start: datetime | None = None\n last_time: datetime | None = None\n for geo, ts in qs:\n loc = extract_location(geo)\n if loc is None:\n continue\n if current_loc is None:\n current_loc = loc\n run_start = ts\n last_time = ts\n continue\n if loc == current_loc:\n last_time = ts\n continue\n # location changed → close previous span\n spans.append((current_loc, run_start, last_time))\n current_loc = loc\n run_start = ts\n last_time = ts\n # close final span\n if current_loc is not None and run_start is not None and last_time is not None:\n spans.append((current_loc, run_start, last_time))\n\n # Coalesce: set each span's end to next span's begin (like SQL LEAD(begin))\n city_start_end_duration = []\n for idx, (loc, begin, end) in enumerate(spans):\n new_end = spans[idx + 1][1] if idx + 1 < len(spans) else end\n duration_sec = (new_end - begin).total_seconds()\n city_start_end_duration.append((loc, begin, new_end, duration_sec))\n\n colors = sns.color_palette(\"Paired\", len(city_start_end_duration)).as_hex()\n\n data = []\n for idx, sted in enumerate(city_start_end_duration):\n data.append(\n {\n \"data\": [sted[3]],\n \"color\": colors[idx],\n \"loc\": sted[0],\n \"start\": sted[1].timestamp(),\n \"end\": sted[2].timestamp(),\n }\n )\n return data\n" }, { "path": "api/tests/__init__.py", "content": "" }, { "path": "api/tests/fixtures/__init__.py", "content": "" }, { "path": "api/tests/fixtures/api_util/captions_json.py", "content": "captions_json = {\n \"places365\": {\n \"attributes\": [\n \"no horizon\",\n \"man made\",\n \"enclosed area\",\n \"cloth\",\n \"natural light\",\n \"wood\",\n \"glass\",\n \"indoor lighting\",\n \"dry\",\n ],\n \"categories\": [\"phone booth\", \"ticket booth\"],\n \"environment\": \"indoor\",\n }\n}\n" }, { "path": "api/tests/fixtures/api_util/expectation.py", "content": "wordcloud_expectation = {\n \"captions\": [\n {\"label\": \"outdoor\", \"y\": 1.9459101490553132},\n {\"label\": \"indoor\", \"y\": 0.6931471805599453},\n {\"label\": \"ticket booth\", \"y\": 0.0},\n {\"label\": \"boardwalk\", \"y\": 0.0},\n {\"label\": \"phone booth\", \"y\": 0.0},\n {\"label\": \"delicatessen\", \"y\": 0.0},\n {\"label\": \"lagoon\", \"y\": 0.0},\n {\"label\": \"tundra\", \"y\": 0.0},\n {\"label\": \"marsh\", \"y\": 0.0},\n {\"label\": \"bakery shop\", \"y\": 0.0},\n {\"label\": \"market outdoor\", \"y\": 0.0},\n {\"label\": \"butchers shop\", \"y\": 0.0},\n {\"label\": \"playground\", \"y\": 0.0},\n {\"label\": \"picnic area\", \"y\": 0.0},\n ],\n \"people\": [],\n \"locations\": [\n {\"label\": \"New South Wales\", \"y\": 1.3862943611198906},\n {\"label\": \"Sydney\", \"y\": 1.3862943611198906},\n {\"label\": \"Australia\", \"y\": 1.3862943611198906},\n {\"label\": \"Maroubra\", \"y\": 1.0986122886681098},\n {\"label\": \"Ladakh\", \"y\": 0.6931471805599453},\n {\"label\": \"Leh\", \"y\": 0.6931471805599453},\n {\"label\": \"Berlin\", \"y\": 0.6931471805599453},\n {\"label\": \"Germany\", \"y\": 0.6931471805599453},\n {\"label\": \"India\", \"y\": 0.6931471805599453},\n {\"label\": \"Lakeshore Road\", \"y\": 0.0},\n {\"label\": \"Shachokol\", \"y\": 0.0},\n {\"label\": \"Kreuzberg\", \"y\": 0.0},\n {\"label\": \"Canada\", \"y\": 0.0},\n {\"label\": \"Bondi Beach\", \"y\": 0.0},\n {\"label\": \"Peterborough County\", \"y\": 0.0},\n {\"label\": \"Lakefield\", \"y\": 0.0},\n {\"label\": \"Main Bazaar\", \"y\": 0.0},\n {\"label\": \"Ontario\", \"y\": 0.0},\n {\"label\": \"Chuchat Yakma\", \"y\": 0.0},\n {\"label\": \"Friedrichshain\", \"y\": 0.0},\n {\"label\": \"Beach Road\", \"y\": 0.0},\n {\"label\": \"Fire Route 47\", \"y\": 0.0},\n ],\n}\n" }, { "path": "api/tests/fixtures/api_util/photos.py", "content": "photos = [\n {\n \"thumbnail_big\": \"thumbnails_big/88070102f4a9a25ba26959b8e1f203a91.webp\",\n \"square_thumbnail\": \"square_thumbnails/88070102f4a9a25ba26959b8e1f203a91.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/88070102f4a9a25ba26959b8e1f203a91.webp\",\n \"added_on\": \"2023-06-16 16:30:47.724635 +00:00\",\n \"exif_gps_lat\": 52.5075472222222,\n \"exif_gps_lon\": 13.4549222222222,\n \"exif_timestamp\": \"2017-08-23 16:13:32.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [13.454922, 52.507547],\n \"features\": [\n {\n \"id\": \"poi.712964618154\",\n \"text\": \"Badehaus - Szimpla Musiksalon\",\n \"type\": \"Feature\",\n \"center\": [13.455101, 52.507538],\n \"context\": [\n {\n \"id\": \"postcode.5156410\",\n \"text\": \"10245\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpUcTQ2\",\n },\n {\n \"id\": \"locality.90794554\",\n \"text\": \"Friedrichshain\",\n \"wikidata\": \"Q317056\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCV2xxT2c\",\n },\n {\n \"id\": \"place.115770\",\n \"text\": \"Berlin\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.455101, 52.507538],\n },\n \"relevance\": 1,\n \"place_name\": \"Badehaus - Szimpla Musiksalon, Revaler Str. 99, Berlin, 10245, Germany\",\n \"place_type\": [\"poi\"],\n \"properties\": {\n \"address\": \"Revaler Str. 99\",\n \"category\": \"bar, pub, alcohol, liquor, beer\",\n \"landmark\": True,\n \"foursquare\": \"4e62924d18a8ce02fce9d584\",\n },\n },\n {\n \"id\": \"postcode.5156410\",\n \"bbox\": [13.445615, 52.486046, 13.491453, 52.514722],\n \"text\": \"10245\",\n \"type\": \"Feature\",\n \"center\": [13.458408, 52.504367],\n \"context\": [\n {\n \"id\": \"locality.90794554\",\n \"text\": \"Friedrichshain\",\n \"wikidata\": \"Q317056\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCV2xxT2c\",\n },\n {\n \"id\": \"place.115770\",\n \"text\": \"Berlin\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.458408, 52.504367],\n },\n \"relevance\": 1,\n \"place_name\": \"10245, Berlin, Germany\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpUcTQ2\"},\n },\n {\n \"id\": \"locality.90794554\",\n \"bbox\": [13.419752, 52.486046, 13.491453, 52.531026],\n \"text\": \"Friedrichshain\",\n \"type\": \"Feature\",\n \"center\": [13.45029, 52.512215],\n \"context\": [\n {\n \"id\": \"place.115770\",\n \"text\": \"Berlin\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [13.45029, 52.512215]},\n \"relevance\": 1,\n \"place_name\": \"Friedrichshain, Berlin, Germany\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q317056\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCV2xxT2c\",\n },\n },\n {\n \"id\": \"place.115770\",\n \"bbox\": [13.08836, 52.338261, 13.761131, 52.675502],\n \"text\": \"Berlin\",\n \"type\": \"Feature\",\n \"center\": [13.3888599, 52.5170365],\n \"context\": [\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.3888599, 52.5170365],\n },\n \"relevance\": 1,\n \"place_name\": \"Berlin, Germany\",\n \"place_type\": [\"region\", \"place\"],\n \"properties\": {\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n },\n {\n \"id\": \"country.8762\",\n \"bbox\": [5.866315, 47.270238, 15.041832, 55.1286491],\n \"text\": \"Germany\",\n \"type\": \"Feature\",\n \"center\": [10.0183432948567, 51.1334813439932],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [10.0183432948567, 51.1334813439932],\n },\n \"relevance\": 1,\n \"place_name\": \"Germany\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Badehaus - Szimpla Musiksalon 10245 Friedrichshain Berlin Germany\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"no horizon\",\n \"man made\",\n \"enclosed area\",\n \"cloth\",\n \"natural light\",\n \"wood\",\n \"glass\",\n \"indoor lighting\",\n \"dry\",\n ],\n \"categories\": [\"phone booth\", \"ticket booth\"],\n \"environment\": \"indoor\",\n }\n },\n \"search_captions\": \"phone booth , ticket booth , indoor\",\n \"search_location\": \"Badehaus - Szimpla Musiksalon 10245 Friedrichshain Berlin Germany\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n 0.02945636957883835,\n 0.0006848685443401337,\n -0.21565294265747070,\n -0.13726805150508880,\n 0.34177422523498535,\n -0.21947097778320312,\n -0.36535099148750305,\n 0.43704378604888916,\n -0.04088297113776207,\n -0.16377356648445130,\n 0.61226522922515870,\n -0.27190640568733215,\n 0.21632727980613708,\n -0.49879187345504760,\n 0.35928872227668760,\n -0.57344657182693480,\n -1.09348952770233150,\n 0.12758482992649078,\n 0.22626115381717682,\n -0.29700523614883423,\n 0.67408502101898190,\n 0.20566113293170930,\n 0.004221141338348389,\n -0.18163934350013733,\n 0.030123591423034668,\n -0.36574870347976685,\n 0.37040638923645020,\n -0.15804094076156616,\n 0.40848651528358460,\n -0.44146019220352173,\n -0.29682993888854980,\n -0.14033758640289307,\n 0.15358698368072510,\n 0.14189875125885010,\n -0.24039962887763977,\n 0.15688608586788177,\n -0.13299843668937683,\n -0.09722766280174255,\n 0.15716867148876190,\n 0.06153672933578491,\n -0.24909989535808563,\n 0.42868158221244810,\n -0.09139122068881989,\n 0.026236191391944885,\n 0.43363174796104430,\n 0.15078209340572357,\n -0.02513679303228855,\n -0.16301257908344270,\n 0.30406942963600160,\n -0.20351555943489075,\n 0.32808005809783936,\n 0.11325813829898834,\n 0.06632904708385468,\n -0.13974805176258087,\n -0.08816416561603546,\n 0.08763262629508972,\n 0.36618012189865110,\n -0.016463160514831543,\n 0.51381558179855350,\n -0.19019323587417603,\n 0.007199913263320923,\n -0.47690716385841370,\n -0.011584945023059845,\n 0.27767223119735720,\n 0.05724636837840080,\n -0.14331746101379395,\n -0.17595694959163666,\n 0.69013822078704830,\n -0.29242870211601260,\n -0.20638096332550050,\n 0.15023268759250640,\n 0.39301073551177980,\n -0.07778433710336685,\n -0.31314268708229065,\n -0.08601596951484680,\n -0.33508300781250000,\n -0.46537816524505615,\n -0.11456200480461120,\n 0.12097512185573578,\n -0.21884436905384064,\n 0.08210694789886475,\n -0.20028553903102875,\n 0.019847501069307327,\n -0.47711223363876340,\n 0.41822510957717896,\n 0.28469321131706240,\n 0.56612819433212280,\n -0.37348130345344543,\n 0.36494156718254090,\n -0.29691773653030396,\n 0.27490460872650146,\n 0.055256836116313934,\n -5.04034137725830100,\n 0.21063046157360077,\n 0.07623146474361420,\n 0.91044491529464720,\n 0.09022644162178040,\n -0.22425013780593872,\n -0.15275251865386963,\n 0.56555438041687010,\n -0.21015779674053192,\n 0.12018989026546478,\n 0.050851717591285706,\n 0.07409352064132690,\n 0.10203989595174790,\n -0.14679333567619324,\n -0.45498293638229370,\n 0.18498001992702484,\n 0.08312227576971054,\n -0.20828101038932800,\n 0.22045168280601501,\n -0.41870164871215820,\n -0.18289424479007720,\n 0.27803760766983030,\n -0.14669309556484222,\n 0.11136791110038757,\n -0.07714871317148209,\n 0.11111782491207123,\n 0.33326071500778200,\n -0.22490462660789490,\n -0.023626238107681274,\n -0.38946658372879030,\n -0.11872470378875732,\n 0.003953278064727783,\n 0.06875754147768020,\n 0.06950961053371430,\n 0.20257902145385742,\n 0.48788213729858400,\n -0.55383354425430300,\n 0.26022326946258545,\n 0.54247814416885380,\n -0.75806522369384770,\n -0.16927206516265870,\n 0.86247527599334720,\n 0.00693318247795105,\n 0.48897692561149597,\n -0.028302762657403946,\n -0.39459863305091860,\n -0.09011757373809814,\n -0.002951778471469879,\n -0.34152612090110780,\n -0.053652018308639526,\n -0.05833066999912262,\n 0.23596245050430298,\n 0.11364882439374924,\n 0.29371896386146545,\n 0.20242361724376678,\n 0.41401028633117676,\n 0.29458260536193850,\n 0.12261603027582169,\n 0.35396143794059753,\n -0.37751555442810060,\n -0.65084457397460940,\n 0.43545067310333250,\n -0.12954968214035034,\n -0.55164968967437740,\n -0.028061915189027786,\n -0.41742083430290220,\n -0.84223127365112300,\n 0.28066438436508180,\n -0.39846715331077576,\n -0.29234075546264650,\n 0.40001532435417175,\n -0.025802582502365112,\n 0.035069286823272705,\n 0.58283793926239010,\n 1.06991338729858400,\n -0.004069690592586994,\n 0.16848376393318176,\n -0.32832053303718567,\n -0.05214332789182663,\n -0.22528168559074402,\n -0.012514784932136536,\n -0.22871030867099762,\n -0.46591621637344360,\n 0.43959569931030273,\n 0.52867388725280760,\n 0.08700185269117355,\n 0.82598596811294560,\n 0.72045898437500000,\n -0.18566657602787018,\n 0.31717744469642640,\n -0.32986947894096375,\n -0.08980216830968857,\n -0.29798981547355650,\n -0.20281662046909332,\n 0.35008898377418520,\n -0.06911731511354446,\n 0.46833136677742004,\n -0.05824550986289978,\n 0.22831164300441742,\n -0.14332316815853120,\n -0.54719811677932740,\n -0.033745840191841125,\n -0.14754259586334229,\n -0.15379694104194640,\n 0.27912741899490356,\n -0.38673168420791626,\n 0.18064787983894348,\n 0.02594844251871109,\n -0.19353626668453217,\n 0.12792934477329254,\n 0.76823699474334720,\n -0.25351685285568240,\n 0.19726955890655518,\n 0.04505273699760437,\n 0.08072985708713531,\n -0.40315085649490356,\n -0.05216634273529053,\n 0.33589959144592285,\n 0.13084968924522400,\n -0.71316510438919070,\n 0.62502771615982060,\n 0.15854838490486145,\n -0.46951773762702940,\n -0.05899134278297424,\n -0.07570817321538925,\n -0.21729907393455505,\n -0.48353233933448790,\n -0.30078506469726560,\n 0.01950731873512268,\n 0.35301312804222107,\n -0.11586478352546692,\n 0.06945395469665527,\n -0.09096579253673553,\n -0.003176487982273102,\n -0.06729926168918610,\n 0.16478994488716125,\n 0.11996459960937500,\n 0.08691802620887756,\n -0.37051337957382200,\n -0.27022105455398560,\n 0.44754734635353090,\n 0.58564084768295290,\n 0.14264307916164398,\n -0.27284860610961914,\n -0.21267735958099365,\n -0.62931954860687260,\n -0.17053475975990295,\n 0.43907707929611206,\n -0.17706312239170074,\n -0.20798830687999725,\n -0.013556711375713348,\n -0.45459657907485960,\n 0.27621361613273620,\n -0.00547829270362854,\n 0.09221091866493225,\n -0.45583301782608030,\n -0.04141671583056450,\n -0.04372096061706543,\n -0.08653303980827332,\n 0.16906698048114777,\n -0.39207679033279420,\n 0.16125959157943726,\n 0.18793570995330810,\n -0.062303997576236725,\n -0.14904721081256866,\n -0.08313135802745819,\n 0.71267020702362060,\n -0.12299664318561554,\n 0.45854219794273376,\n 0.37172207236289980,\n 0.23576152324676514,\n -0.96158236265182500,\n -0.16355091333389282,\n 0.08433900773525238,\n -0.26110875606536865,\n 0.41418594121932983,\n -0.00952976569533348,\n -0.31196358799934387,\n 0.08939869701862335,\n 0.08166363835334778,\n 0.27331307530403137,\n -0.23260714113712310,\n -0.11067157238721848,\n 0.07786403596401215,\n 0.15424127876758575,\n -0.15310212969779968,\n -0.57094782590866090,\n 0.08447279036045074,\n 0.09531763195991516,\n -0.16973701119422913,\n -0.39517831802368164,\n 0.48301315307617190,\n -0.49167299270629883,\n 0.16578084230422974,\n 0.012142054736614227,\n 0.38737818598747253,\n -0.42219209671020510,\n 0.038790080696344376,\n 0.14427416026592255,\n 0.51845633983612060,\n 0.30745857954025270,\n 0.19423109292984010,\n -0.29949772357940674,\n -0.22886085510253906,\n -0.18502810597419740,\n 0.58494806289672850,\n -0.05995021387934685,\n 0.027786150574684143,\n 0.16225285828113556,\n -0.21960149705410004,\n -0.19987022876739502,\n 0.17615880072116852,\n 0.12230071425437927,\n 0.78366780281066900,\n 0.07176887989044190,\n 0.34315368533134460,\n 0.32958623766899110,\n 0.046860724687576294,\n -0.37855774164199830,\n 0.10880301892757416,\n 0.86260342597961430,\n -0.25351122021675110,\n -0.33743318915367126,\n -0.000004470348358154297,\n 0.10894200950860977,\n -0.24402043223381042,\n 0.041552767157554626,\n 1.11131787300109860,\n 0.19667577743530273,\n 0.09909820556640625,\n -0.22913703322410583,\n 0.07402139902114868,\n -0.51243948936462400,\n -0.011785134673118591,\n 0.11913090944290161,\n 0.10973167419433594,\n 0.14159473776817322,\n -0.25530740618705750,\n -0.10128769278526306,\n -0.018086418509483337,\n 0.23918014764785767,\n -0.0025010444223880768,\n 0.028242304921150208,\n -0.10461166501045227,\n 0.31454890966415405,\n -0.21300530433654785,\n -0.67316675186157230,\n -0.28894531726837160,\n 0.15850226581096650,\n -0.25393971800804140,\n 0.015249133110046387,\n 0.17902487516403198,\n 0.03674941509962082,\n -0.40653368830680847,\n 0.06983637809753418,\n 0.58390831947326660,\n -0.35204166173934937,\n -0.23273511230945587,\n -0.50346362590789800,\n -0.18698497116565704,\n -0.13735061883926392,\n -0.029784053564071655,\n -0.23463270068168640,\n 0.04800243675708771,\n 0.21194185316562653,\n 2.50413703918457030,\n 0.014399580657482147,\n 0.13682425022125244,\n -0.44549101591110230,\n -0.05332183837890625,\n 0.26232331991195680,\n 0.48867249488830566,\n 0.90782088041305540,\n -0.04284442216157913,\n 0.16441944241523743,\n -0.64545953273773190,\n 0.0021270206198096275,\n 0.57723963260650630,\n -0.16735523939132690,\n -0.03681857883930206,\n -0.23049485683441162,\n -0.16818353533744812,\n 0.48354497551918030,\n -0.34307366609573364,\n 0.90225481986999510,\n -0.08604586869478226,\n -0.80853521823883060,\n 0.07412493973970413,\n 0.38500127196311950,\n -0.53182345628738400,\n -0.37833765149116516,\n -0.27583479881286620,\n 0.18323482573032380,\n -0.57118958234786990,\n 0.15203747153282166,\n -0.45285981893539430,\n 0.14797788858413696,\n 0.92780256271362300,\n -0.99070614576339720,\n 0.78525125980377200,\n -0.54372054338455200,\n 0.27046817541122437,\n 0.05927395448088646,\n 0.29292181134223940,\n 0.17652636766433716,\n -0.28721928596496580,\n -0.11287573724985123,\n 0.054300934076309204,\n -0.20183458924293518,\n -0.06938034296035767,\n -0.02237860858440399,\n -0.28468230366706850,\n 0.27039328217506410,\n -0.44048523902893066,\n -0.10478976368904114,\n 0.22237929701805115,\n 0.61014431715011600,\n 0.22426150739192963,\n -0.53682762384414670,\n 0.17780613899230957,\n 0.10927680134773254,\n -0.08265303075313568,\n 0.08433099091053009,\n 0.22053673863410950,\n -0.15935529768466950,\n 0.02448994666337967,\n -0.15990000963211060,\n -0.10862783342599869,\n -0.55222094058990480,\n -0.18120177090168000,\n -0.18480616807937622,\n -0.40381407737731934,\n 0.41113349795341490,\n 0.45374616980552673,\n 0.12371205538511276,\n 0.54327690601348880,\n -1.32808542251586910,\n 0.06879405677318573,\n -0.32771402597427370,\n 0.30996549129486084,\n 0.73747062683105470,\n 0.0026262253522872925,\n -0.07966820895671844,\n -0.43908175826072693,\n -0.57706052064895630,\n -0.07506740093231201,\n -0.03822087496519089,\n 0.14753995835781097,\n 0.26234555244445800,\n 0.37969052791595460,\n -0.03944581747055054,\n -0.21361999213695526,\n 0.30636596679687500,\n 0.36208370327949524,\n -0.51721900701522830,\n -0.72269493341445920,\n 0.12646321952342987,\n 0.30191451311111450,\n 0.88483667373657230,\n 0.13586276769638062,\n -0.47170540690422060,\n -0.22053156793117523,\n -0.32388603687286377,\n 0.28210371732711790,\n -0.14361841976642610,\n 0.35813957452774050,\n -0.71662276983261110,\n -0.36910200119018555,\n -0.50887799263000490,\n 0.18198430538177490,\n -0.12547180056571960,\n 0.019948184490203857,\n -0.027965955436229706,\n 0.23368665575981140,\n 0.08205719292163849,\n 0.03229612857103348,\n -0.37393200397491455,\n 0.09985578060150146,\n 0.16484981775283813,\n -0.03449898585677147,\n -0.20320037007331848,\n 0.17666932940483093,\n -0.10235498100519180,\n -0.18969213962554932,\n -0.06653118133544922,\n 0.32699528336524963,\n -0.41083279252052307,\n 0.17249025404453278,\n -0.71300625801086430,\n -0.40383389592170715,\n -0.012773919850587845,\n -0.04669109359383583,\n -0.16580447554588318,\n -0.034973472356796265,\n 0.18406766653060913,\n 0.32842925190925600,\n -0.43452721834182740,\n 0.07465618103742600,\n -0.054404426366090775,\n 0.07043577730655670,\n 0.34492620825767517,\n -0.55067378282547000,\n -0.35782644152641296,\n 0.22946852445602417,\n 0.13374575972557068,\n 0.23378580808639526,\n -0.18907654285430908,\n -0.20370183885097504,\n -0.63759642839431760,\n 0.29174658656120300,\n -0.20586502552032470,\n 0.08419455587863922,\n 0.96353048086166380,\n -0.023577764630317688,\n 0.34594947099685670,\n -0.07180836796760559,\n -0.11015791445970535,\n -0.10569672286510468,\n 0.29831579327583313,\n 0.21254715323448180,\n ],\n \"clip_embeddings_magnitude\": 9.564716339111328,\n \"aspect_ratio\": 0.75,\n \"rating\": 0,\n \"dominant_color\": \"[145, 83, 48]\",\n \"video_length\": None,\n \"in_trashcan\": False,\n \"removed\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 4032,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"1/100\",\n \"size\": 7194067,\n \"subjectDistance\": 0.614,\n \"width\": 3024,\n \"main_file_id\": \"88070102f4a9a25ba26959b8e1f203a91\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/cac6402c7ff192e6fa96fee50ba24fd81.webp\",\n \"square_thumbnail\": \"square_thumbnails/cac6402c7ff192e6fa96fee50ba24fd81.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/cac6402c7ff192e6fa96fee50ba24fd81.webp\",\n \"added_on\": \"2023-06-16 16:30:44.284649 +00:00\",\n \"exif_gps_lat\": 34.1620972222222,\n \"exif_gps_lon\": 77.5858416666667,\n \"exif_timestamp\": \"2017-08-09 23:04:03.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [77.585842, 34.162097],\n \"features\": [\n {\n \"id\": \"address.15438142488176\",\n \"text\": \"Main Bazaar\",\n \"type\": \"Feature\",\n \"center\": [77.585527, 34.1622089],\n \"context\": [\n {\n \"id\": \"postcode.13053547\",\n \"text\": \"194101\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzp4eTVy\",\n },\n {\n \"id\": \"locality.997100139\",\n \"text\": \"Chuchat Yakma\",\n \"wikidata\": \"Q24909733\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpPMjZLYXc\",\n },\n {\n \"id\": \"place.25135211\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.585527, 34.1622089],\n },\n \"relevance\": 1,\n \"place_name\": \"Main Bazaar ، 194101 Leh، India\",\n \"place_type\": [\"address\"],\n \"properties\": {\"accuracy\": \"street\"},\n },\n {\n \"id\": \"postcode.13053547\",\n \"bbox\": [77.107033, 33.67805, 77.812822, 34.511912],\n \"text\": \"194101\",\n \"type\": \"Feature\",\n \"center\": [77.572696, 34.199537],\n \"context\": [\n {\n \"id\": \"locality.997100139\",\n \"text\": \"Chuchat Yakma\",\n \"wikidata\": \"Q24909733\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpPMjZLYXc\",\n },\n {\n \"id\": \"place.25135211\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.572696, 34.199537],\n },\n \"relevance\": 1,\n \"place_name\": \"194101, Leh, Ladakh, India\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzp4eTVy\"},\n },\n {\n \"id\": \"locality.997100139\",\n \"bbox\": [76.626529011, 33.347769, 78.047440762, 34.538001],\n \"text\": \"Chuchat Yakma\",\n \"type\": \"Feature\",\n \"center\": [77.602518, 34.07534],\n \"context\": [\n {\n \"id\": \"place.25135211\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [77.602518, 34.07534]},\n \"relevance\": 1,\n \"place_name\": \"Chuchat Yakma, Leh, Leh, Ladakh, India\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q24909733\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpPMjZLYXc\",\n },\n },\n {\n \"id\": \"place.25135211\",\n \"bbox\": [77.107033, 32.33574, 79.305839, 35.522256],\n \"text\": \"Leh\",\n \"type\": \"Feature\",\n \"center\": [77.584813, 34.164203],\n \"context\": [\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.584813, 34.164203],\n },\n \"relevance\": 1,\n \"place_name\": \"Leh, Ladakh, India\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n },\n {\n \"id\": \"region.222315\",\n \"bbox\": [75.306629, 32.33574, 79.305851, 35.673315],\n \"text\": \"Ladakh\",\n \"type\": \"Feature\",\n \"center\": [77.27783203125, 34.0071350643588],\n \"context\": [\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.27783203125, 34.0071350643588],\n },\n \"relevance\": 1,\n \"place_name\": \"Ladakh, India\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n },\n {\n \"id\": \"country.8811\",\n \"bbox\": [68.1152344, 6.6718373, 97.395359, 35.673315],\n \"text\": \"India\",\n \"type\": \"Feature\",\n \"center\": [78.476681027237, 22.1991660760527],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [78.476681027237, 22.1991660760527],\n },\n \"relevance\": 1,\n \"place_name\": \"India\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Main Bazaar 194101 Chuchat Yakma Leh Ladakh India\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"no horizon\",\n \"cloth\",\n \"enclosed area\",\n \"natural light\",\n \"man made\",\n \"rugged scene\",\n \"dry\",\n \"shopping\",\n \"working\",\n ],\n \"categories\": [\n \"butchers shop\",\n \"bakery shop\",\n \"delicatessen\",\n \"market outdoor\",\n ],\n \"environment\": \"indoor\",\n }\n },\n \"search_captions\": \"butchers shop , bakery shop , delicatessen , market outdoor , indoor\",\n \"search_location\": \"Main Bazaar 194101 Chuchat Yakma Leh Ladakh India\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n -0.36275115609169006,\n 0.04232349991798401,\n 0.16003805398941040,\n -0.15414649248123170,\n 0.07362174987792969,\n 0.30168089270591736,\n 0.29503446817398070,\n 0.52876400947570800,\n 0.03624974191188812,\n 0.57748210430145260,\n 0.06661064177751541,\n -0.08749669790267944,\n -0.43980333209037780,\n -0.45719712972640990,\n 0.20980522036552430,\n 0.31552031636238100,\n 0.28292679786682130,\n -0.23881813883781433,\n 0.26318275928497314,\n -0.25081199407577515,\n -1.56406641006469730,\n -0.20272035896778107,\n -0.30980134010314940,\n -0.13830965757369995,\n -0.65654301643371580,\n 0.31328544020652770,\n 0.39124441146850586,\n 0.14648842811584473,\n 0.10790233314037323,\n -0.0011418797075748444,\n 0.29148331284523010,\n 0.39446878433227540,\n 0.15861868858337402,\n -0.01094374805688858,\n 0.69656980037689210,\n 0.53402501344680790,\n -0.19458736479282380,\n -0.33765941858291626,\n 0.20578813552856445,\n 1.02007591724395750,\n -0.27630817890167236,\n -0.055134907364845276,\n 0.10728104412555695,\n 0.19857768714427948,\n -0.051949791610240936,\n -0.95591288805007930,\n -0.016068458557128906,\n 0.14163839817047120,\n 0.07818788290023804,\n 0.30343624949455260,\n -0.00487295538187027,\n -0.03668808937072754,\n 0.23092812299728394,\n 0.22601000964641570,\n -0.17142324149608612,\n 0.19191186130046844,\n 0.24479267001152039,\n 0.75923907756805420,\n 0.52936017513275150,\n 0.59186935424804690,\n -0.60621523857116700,\n 0.32727080583572390,\n -0.11824478209018707,\n -0.05898610129952431,\n -0.012164704501628876,\n -0.10524882376194000,\n -0.64567971229553220,\n 1.40957057476043700,\n -0.30128875374794006,\n -0.52295249700546260,\n 0.053494296967983246,\n -0.34456601738929750,\n -0.22763675451278687,\n -0.07936172187328339,\n -0.61760181188583370,\n -0.18934124708175660,\n -0.04926346242427826,\n 0.07053080201148987,\n -0.32558596134185790,\n -0.61692816019058230,\n 0.25635927915573120,\n -0.40595006942749023,\n -0.45488622784614563,\n -1.27675628662109380,\n 0.23719409108161926,\n -0.24088490009307860,\n -0.042445749044418335,\n -0.0018720626831054688,\n 1.05558359622955320,\n -0.49247956275939940,\n 0.17333477735519410,\n 0.035649579018354416,\n -5.73797512054443400,\n 1.30945181846618650,\n 0.51552683115005490,\n 0.15089595317840576,\n 0.35859110951423645,\n -0.04071284830570221,\n -0.86825460195541380,\n 1.31599712371826170,\n -0.09168644249439240,\n 0.63117879629135130,\n -0.17361941933631897,\n 0.10322868078947067,\n -0.60555946826934810,\n 0.23382598161697388,\n -0.82185673713684080,\n -0.10401326417922974,\n 0.13238205015659332,\n -0.03597678989171982,\n -0.51136314868927000,\n 0.15768136084079742,\n -0.23854430019855500,\n -0.14175727963447570,\n -0.11781750619411469,\n 0.38926169276237490,\n -0.31990337371826170,\n 0.24176508188247680,\n 0.59939318895339970,\n -0.72715508937835690,\n 0.39353600144386290,\n -0.38137459754943850,\n 0.20685830712318420,\n -0.38994306325912476,\n 0.05647854506969452,\n -0.23935276269912720,\n -0.22889091074466705,\n -0.019344180822372437,\n 0.040057696402072906,\n -0.24383866786956787,\n -0.39092701673507690,\n -0.018549658358097076,\n 0.51890051364898680,\n 0.86515939235687260,\n 0.28614503145217896,\n 0.64386940002441410,\n 0.31628671288490295,\n -0.40825888514518740,\n -0.33183339238166810,\n 0.16869042813777924,\n 0.010168105363845825,\n 0.30278283357620240,\n -0.60569459199905400,\n -0.20262216031551360,\n -1.09207832813262940,\n -0.027563804760575294,\n 0.06484303623437881,\n 0.27420419454574585,\n -0.49345779418945310,\n 0.10804055631160736,\n -0.14800167083740234,\n 0.08484844118356705,\n -0.78326535224914550,\n -0.006926223635673523,\n -0.25266957283020020,\n -0.26330786943435670,\n 0.89378148317337040,\n -0.64368873834609990,\n 0.25188437104225160,\n 0.04949730634689331,\n -0.23263952136039734,\n -0.0038242843002080917,\n 0.24761386215686798,\n 0.21502655744552612,\n 0.27725243568420410,\n -0.03591444715857506,\n 0.11199191957712173,\n 0.35744479298591614,\n 0.15654230117797852,\n -0.11716540157794952,\n -0.49658292531967163,\n -0.12044863402843475,\n -0.59336757659912110,\n -0.04827989637851715,\n 0.27895766496658325,\n 0.60082894563674930,\n -0.44568234682083130,\n 0.18149539828300476,\n 0.84179604053497310,\n -0.20165166258811950,\n 0.15945611894130707,\n -0.19363774359226227,\n -0.38449636101722720,\n -0.38157621026039124,\n 0.23142817616462708,\n -0.32686632871627810,\n 0.32904222607612610,\n -0.62547576427459720,\n -0.17987522482872010,\n 0.13854908943176270,\n 0.28109687566757200,\n -0.66307181119918820,\n -0.10197299718856812,\n 0.23179894685745240,\n 0.14371933043003082,\n -0.015187196433544159,\n -0.005701176822185516,\n -0.009048223495483398,\n -0.36947503685951233,\n -0.08419185131788254,\n 0.18359073996543884,\n 0.36397165060043335,\n -0.18554688990116120,\n 0.06282243132591248,\n 0.12127010524272919,\n 0.10142311453819275,\n -0.75937509536743160,\n 0.14368136227130890,\n -0.15232974290847778,\n 0.030483879148960114,\n -0.20716962218284607,\n 0.09505847096443176,\n -0.03633452206850052,\n 0.11635903269052505,\n 0.19411554932594300,\n -0.34413999319076540,\n 0.09673729538917542,\n 0.15466451644897460,\n -0.11886352300643921,\n -0.31654387712478640,\n 0.023902088403701782,\n 0.59250128269195560,\n -0.03467556834220886,\n -0.19663128256797790,\n 0.07225932180881500,\n -0.20673781633377075,\n 0.10626789927482605,\n 0.27759736776351930,\n 0.14237537980079650,\n -0.06108829379081726,\n -0.14290234446525574,\n -0.034046679735183716,\n -0.23872807621955872,\n -0.30120781064033510,\n 0.53876233100891110,\n 0.47913902997970580,\n 0.15108020603656770,\n -0.21000367403030396,\n -0.09006668627262115,\n -0.12558355927467346,\n -0.29648718237876890,\n -0.09471682459115982,\n 0.10165409743785858,\n -0.04843531921505928,\n -0.20445811748504640,\n -0.64695239067077640,\n 0.34345996379852295,\n 0.25094231963157654,\n 0.28040134906768800,\n 0.26930731534957886,\n -0.18992258608341217,\n -0.31795889139175415,\n 0.10275565087795258,\n -0.31493186950683594,\n 0.05224027484655380,\n 0.03900603950023651,\n -0.15962079167366028,\n -0.40141177177429200,\n 1.50395190715789800,\n -0.18131579458713531,\n 0.25262355804443360,\n 0.56724452972412110,\n 0.19896474480628967,\n 0.22113317251205444,\n -0.37682905793190000,\n -0.34762057662010193,\n -0.36291345953941345,\n 0.38737395405769350,\n 0.39863935112953186,\n 0.06930093467235565,\n -0.07484766840934753,\n 0.26550960540771484,\n -0.37431126832962036,\n 0.06163658946752548,\n -0.16546487808227540,\n -0.16572664678096770,\n -0.41679629683494570,\n -0.22878777980804443,\n 0.17536523938179016,\n 0.43185859918594360,\n -0.18755021691322327,\n 0.57233977317810060,\n 0.34548804163932800,\n 0.26037001609802246,\n 0.57140541076660160,\n 1.03332614898681640,\n -1.69704651832580570,\n 0.44388985633850100,\n 0.08646445721387863,\n -0.32414677739143370,\n -0.39306485652923584,\n 0.21713417768478394,\n -0.38696998357772827,\n 0.16778808832168580,\n -0.05270355939865112,\n -0.22387394309043884,\n 0.07469445466995239,\n 0.04779075086116791,\n 0.30713665485382080,\n -0.33245295286178590,\n -0.26004213094711304,\n -0.12230075895786285,\n -0.65894758701324460,\n 0.06937650591135025,\n 0.11567105352878570,\n 0.38836821913719180,\n -0.07238084077835083,\n 0.58317494392395020,\n 0.71717137098312380,\n 0.10365724563598633,\n -0.58237200975418090,\n 0.01732088252902031,\n 0.86498779058456420,\n -0.04113391041755676,\n -0.09420709311962128,\n 0.19938360154628754,\n 0.004110394045710564,\n 0.0035112202167510986,\n 0.32821851968765260,\n 0.20233154296875000,\n 0.10732693225145340,\n 1.30262577533721920,\n -0.25104320049285890,\n -0.09294586628675461,\n 0.07033525407314300,\n 0.26292678713798523,\n -0.10476322472095490,\n 0.49015024304389954,\n -0.38899934291839600,\n -0.057010307908058167,\n -0.19918194413185120,\n -0.23839278519153595,\n 0.34912475943565370,\n -0.38353130221366880,\n 0.034570418298244476,\n -0.10393957048654556,\n -0.19936946034431458,\n 0.30964389443397520,\n -0.012369591742753983,\n -0.14967474341392517,\n -0.11177364736795425,\n -0.19989341497421265,\n 0.62171781063079830,\n 0.20537722110748290,\n 0.16507479548454285,\n -0.59005945920944210,\n -0.11848224699497223,\n 0.00022661685943603516,\n 0.03684444725513458,\n 0.06734884530305862,\n -0.06063830852508545,\n 0.06855276226997375,\n 0.38286936283111570,\n -0.07564809173345566,\n 0.11344917118549347,\n -0.42775198817253113,\n -0.23143202066421510,\n -0.13804715871810913,\n 0.08444565534591675,\n -0.17835569381713867,\n -0.56565302610397340,\n -0.04202693700790405,\n 0.09906928986310959,\n 0.78766095638275150,\n 0.38724774122238160,\n 0.22494032979011536,\n -0.09758158773183823,\n 0.30130439996719360,\n 0.08548129349946976,\n 0.012833205051720142,\n -0.33451801538467410,\n -0.02345825731754303,\n -0.11464481055736542,\n 0.20209941267967224,\n 0.13876622915267944,\n -0.09112481772899628,\n 0.59946906566619870,\n -0.31222972273826600,\n -0.15563198924064636,\n -0.019779425114393234,\n -0.18419799208641052,\n -0.36252117156982420,\n -0.66804647445678710,\n 0.07264155894517899,\n 0.14690190553665160,\n 0.58527380228042600,\n 0.10589228570461273,\n 0.15317067503929138,\n -0.14340689778327942,\n 0.11329543590545654,\n 0.21681559085845947,\n -0.01961892843246460,\n 0.22408972680568695,\n 0.40270829200744630,\n -0.28345388174057007,\n 0.001580655574798584,\n 0.29231256246566770,\n -0.31483244895935060,\n -0.66095465421676640,\n 0.12513107061386108,\n 0.25342079997062683,\n -0.35947406291961670,\n 0.10373535752296448,\n -0.14622645080089570,\n 0.03869847208261490,\n -0.13646872341632843,\n -0.054877884685993195,\n -0.29587787389755250,\n 0.10567575693130493,\n -0.27623349428176880,\n -0.16489294171333313,\n 0.03129418194293976,\n -0.36633712053298950,\n -0.30330657958984375,\n -0.10047523677349090,\n 0.38868999481201170,\n 0.18341153860092163,\n -0.42602014541625977,\n -0.42138785123825073,\n 0.16399800777435303,\n -0.13668435811996460,\n 0.08918181061744690,\n 0.18008530139923096,\n 0.34570962190628050,\n 0.13715223968029022,\n -0.35308316349983215,\n 0.017223667353391647,\n 0.16259710490703583,\n -2.01324987411499020,\n 0.35719367861747740,\n 0.05423700809478760,\n 0.64791858196258540,\n 1.41284525394439700,\n -0.025331489741802216,\n 0.03643947094678879,\n -0.07986896485090256,\n -0.45607382059097290,\n -0.36353543400764465,\n -0.022743068635463715,\n 0.41851943731307983,\n 0.05244317650794983,\n 0.17263151705265045,\n 0.07288816571235657,\n 0.30261635780334470,\n -0.38646593689918520,\n -0.054808780550956726,\n -0.28826522827148440,\n -0.01994667947292328,\n -0.48321542143821716,\n -0.21889880299568176,\n -0.26040261983871460,\n -0.52937489748001100,\n -0.041442275047302246,\n -1.15707290172576900,\n 0.22746005654335022,\n -0.12931922078132630,\n 0.35249620676040650,\n 0.27543902397155760,\n -0.39276057481765747,\n -0.42217552661895750,\n -0.22715577483177185,\n 0.004897281527519226,\n -0.31350314617156980,\n -0.45807397365570070,\n 0.25268691778182983,\n 0.04663296043872833,\n -0.59933739900588990,\n 0.15925574302673340,\n -0.08339263498783112,\n 0.18594521284103394,\n -0.13940119743347168,\n 0.30508360266685486,\n -0.24364741146564484,\n -0.64189815521240230,\n -0.02178923785686493,\n -0.22042772173881530,\n 0.34853425621986390,\n 0.46653327345848083,\n -0.03486161679029465,\n 0.50477695465087890,\n -0.10376831889152527,\n -0.14050556719303130,\n -0.08073642104864120,\n 0.06925593316555023,\n 0.21366350352764130,\n 0.24056504666805267,\n 0.12795145809650420,\n 0.16419601440429688,\n 0.06756377220153809,\n -0.18162062764167786,\n 0.33428809046745300,\n 0.06203627586364746,\n 0.63255858421325680,\n -0.51533907651901250,\n 0.13819636404514313,\n 0.07493585348129272,\n -0.19676387310028076,\n 0.30928391218185425,\n -0.47425344586372375,\n 0.09209989756345749,\n 0.44058740139007570,\n 0.24711114168167114,\n -0.32648178935050964,\n 0.40268272161483765,\n 0.11267729103565216,\n 0.17878434062004090,\n -0.52805340290069580,\n 0.46472996473312380,\n -0.47340789437294006,\n 0.62205803394317630,\n -0.28207665681838990,\n 0.04906800761818886,\n ],\n \"clip_embeddings_magnitude\": 10.556536674499512,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[216, 130, 71]\",\n \"video_length\": None,\n \"removed\": False,\n \"in_trashcan\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"1/1000\",\n \"size\": 7487773,\n \"subjectDistance\": 0.265,\n \"width\": 4032,\n \"main_file_id\": \"cac6402c7ff192e6fa96fee50ba24fd81\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/1f522608a263a88ea72010fe4f08f3071.webp\",\n \"square_thumbnail\": \"square_thumbnails/1f522608a263a88ea72010fe4f08f3071.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/1f522608a263a88ea72010fe4f08f3071.webp\",\n \"added_on\": \"2023-06-16 16:30:51.051404 +00:00\",\n \"exif_gps_lat\": -33.941975,\n \"exif_gps_lon\": 151.265088888889,\n \"exif_timestamp\": \"2017-10-17 10:13:13.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [151.265089, -33.941975],\n \"features\": [\n {\n \"id\": \"poi.721554610670\",\n \"text\": \"Mistral Point\",\n \"type\": \"Feature\",\n \"center\": [151.26523, -33.941658],\n \"context\": [\n {\n \"id\": \"postcode.503310\",\n \"text\": \"2035\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCNjRP\",\n },\n {\n \"id\": \"locality.269412878\",\n \"text\": \"Maroubra\",\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.26523, -33.941658],\n },\n \"relevance\": 1,\n \"place_name\": \"Mistral Point, Sydney, New South Wales 2035, Australia\",\n \"place_type\": [\"poi\"],\n \"properties\": {\n \"category\": \"historic site, historic\",\n \"landmark\": True,\n \"foursquare\": \"5958b8724420d86b1808f7bd\",\n },\n },\n {\n \"id\": \"postcode.503310\",\n \"bbox\": [151.205214, -33.958015, 151.265693, -33.931801],\n \"text\": \"2035\",\n \"type\": \"Feature\",\n \"center\": [151.242438, -33.942906],\n \"context\": [\n {\n \"id\": \"locality.269412878\",\n \"text\": \"Maroubra\",\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.242438, -33.942906],\n },\n \"relevance\": 1,\n \"place_name\": \"2035, Maroubra, New South Wales, Australia\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpCNjRP\"},\n },\n {\n \"id\": \"locality.269412878\",\n \"bbox\": [151.226804557, -33.958002321, 151.292531935, -33.93283624],\n \"text\": \"Maroubra\",\n \"type\": \"Feature\",\n \"center\": [151.2575, -33.9475],\n \"context\": [\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [151.2575, -33.9475]},\n \"relevance\": 1,\n \"place_name\": \"Maroubra, New South Wales, Australia\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n },\n {\n \"id\": \"place.24496142\",\n \"bbox\": [150.520934139, -34.11717528, 151.369884128, -33.562644328],\n \"text\": \"Sydney\",\n \"type\": \"Feature\",\n \"center\": [151.216454, -33.854816],\n \"context\": [\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.216454, -33.854816],\n },\n \"relevance\": 1,\n \"place_name\": \"Sydney, New South Wales, Australia\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n },\n {\n \"id\": \"region.33806\",\n \"bbox\": [140.999265, -37.5097258, 159.200456, -28.1370359],\n \"text\": \"New South Wales\",\n \"type\": \"Feature\",\n \"center\": [147.014694071448, -32.168971672412],\n \"context\": [\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [147.014694071448, -32.168971672412],\n },\n \"relevance\": 1,\n \"place_name\": \"New South Wales, Australia\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n },\n {\n \"id\": \"country.8718\",\n \"bbox\": [112.8256904, -54.8327658, 159.200456, -9.0436707],\n \"text\": \"Australia\",\n \"type\": \"Feature\",\n \"center\": [134.489562606981, -25.7349684916223],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [134.489562606981, -25.7349684916223],\n },\n \"relevance\": 1,\n \"place_name\": \"Australia\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Mistral Point 2035 Maroubra Sydney New South Wales Australia\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"natural light\",\n \"open area\",\n \"cloth\",\n \"sunny\",\n \"rugged scene\",\n \"man made\",\n \"no horizon\",\n \"dry\",\n \"climbing\",\n ],\n \"categories\": [],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"outdoor\",\n \"search_location\": \"Mistral Point 2035 Maroubra Sydney New South Wales Australia\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n -0.07165607810020447,\n 0.27777394652366640,\n -0.15911960601806640,\n -0.008188512176275253,\n -0.009010881185531616,\n 0.034435927867889404,\n -0.03819745033979416,\n 0.64109969139099120,\n 0.26547160744667053,\n 0.12205618619918823,\n 0.29898315668106080,\n -0.10334454476833344,\n 0.48246607184410095,\n -0.35823887586593630,\n -0.024076826870441437,\n -0.39234489202499390,\n -1.93601369857788090,\n -0.038323208689689636,\n 0.26534727215766907,\n -0.31557315587997437,\n 0.42033576965332030,\n 0.10789811611175537,\n 0.25757673382759094,\n -0.69643688201904300,\n 0.02039242535829544,\n 0.61391353607177730,\n -0.14854159951210022,\n -0.02548396587371826,\n 0.09138785302639008,\n 0.01601754128932953,\n -0.16707235574722290,\n 0.18399655818939210,\n -0.35197073221206665,\n -0.67160075902938840,\n -0.25247159600257874,\n 0.025167599320411682,\n -0.03638566657900810,\n 0.30362653732299805,\n 0.54022008180618290,\n -0.18761183321475983,\n -0.54178249835968020,\n 0.14752727746963500,\n 0.14671368896961212,\n -0.26148569583892820,\n 0.16243551671504974,\n -1.18054628372192380,\n 0.53845894336700440,\n -0.16278918087482452,\n -0.22392278909683228,\n -0.31029137969017030,\n 0.10120061784982681,\n 0.36866247653961180,\n 0.68476313352584840,\n -0.32936185598373413,\n 0.42605328559875490,\n 0.29965081810951233,\n -0.09844471514225006,\n -0.25767421722412110,\n -0.34776532649993896,\n -0.34964543581008910,\n -0.35941550135612490,\n -0.04731412231922150,\n -0.12926623225212097,\n 0.11695092916488647,\n -0.32863533496856690,\n -0.19086918234825134,\n -0.16123855113983154,\n 1.26160860061645500,\n -0.38261440396308900,\n -0.038163766264915466,\n -0.17712144553661346,\n 0.07694791257381439,\n -0.38384479284286500,\n -0.39457264542579650,\n -0.024550501257181168,\n -0.30032104253768920,\n -0.48474478721618650,\n 0.17176222801208496,\n -0.35844194889068604,\n -0.22569467127323150,\n -0.04710277169942856,\n 0.26759219169616700,\n 0.18299442529678345,\n 0.42168390750885010,\n 0.27530056238174440,\n 0.08983490616083145,\n 0.56285107135772700,\n -0.36590436100959780,\n -0.21663343906402588,\n -0.23292037844657898,\n 0.28227549791336060,\n -0.03927003592252731,\n -6.69503688812255900,\n 0.40044450759887695,\n 0.17168624699115753,\n 0.16189746558666230,\n -0.21396432816982270,\n -0.18209101259708405,\n -0.14555342495441437,\n -1.11969792842865000,\n 0.26553612947463990,\n -0.43248271942138670,\n -0.23711906373500824,\n 0.13495093584060670,\n -0.57171511650085450,\n 0.31623297929763794,\n -0.26300746202468870,\n 0.11482997238636017,\n -0.04145537316799164,\n -0.012881413102149963,\n 0.38327059149742126,\n -0.28872770071029663,\n 0.13417448103427887,\n 0.0004943050444126129,\n -0.48013591766357420,\n 0.15905582904815674,\n -0.17036144435405730,\n 0.11494286358356476,\n 0.38910707831382750,\n -0.04960992932319641,\n 0.0068108439445495605,\n -0.39681759476661680,\n 0.05528683960437775,\n 0.25615781545639040,\n 0.16935455799102783,\n -0.15641900897026062,\n 0.030953165143728256,\n -0.09216502308845520,\n -0.51438409090042110,\n 0.20902346074581146,\n 0.09113435447216034,\n -0.28546446561813354,\n 0.12450934201478958,\n 0.90055423974990840,\n -0.30295214056968690,\n 0.04577522724866867,\n 0.23674716055393220,\n 0.19620284438133240,\n -0.14195847511291504,\n 0.027747787535190582,\n 0.24356174468994140,\n -0.24453750252723694,\n -0.13842593133449554,\n -0.04814389348030090,\n -0.29600149393081665,\n 0.33813029527664185,\n -0.40565192699432373,\n 1.04664814472198490,\n -0.12125436961650848,\n -0.04555452987551689,\n -0.32212057709693910,\n 0.19244483113288880,\n -0.17625807225704193,\n -0.11717353016138077,\n 0.52924054861068730,\n -0.91078692674636840,\n 0.61408323049545290,\n -0.015624964609742165,\n 0.005238614976406097,\n 0.49075758457183840,\n -0.62067019939422610,\n 0.30432268977165220,\n -0.032360561192035675,\n 0.30186992883682250,\n -0.41565680503845215,\n -0.36446920037269590,\n 0.94007527828216550,\n 0.70317226648330690,\n 0.23549987375736237,\n -0.39121294021606445,\n -0.17178057134151460,\n -0.042171500623226166,\n 0.05205693840980530,\n -0.12189273536205292,\n -0.19439676403999330,\n -0.12377274036407471,\n -0.34452289342880250,\n 0.02267320826649666,\n -0.68486988544464110,\n 0.55518913269042970,\n 0.44782432913780210,\n 0.029707305133342743,\n -0.07340695708990097,\n 0.25972485542297363,\n -0.25381219387054443,\n 0.19427055120468140,\n 0.29389485716819763,\n -0.50578629970550540,\n 0.15614330768585205,\n 0.31318914890289307,\n -0.08632530272006989,\n -0.23409178853034973,\n 0.05476500093936920,\n 0.44156715273857117,\n -0.22440673410892487,\n 0.023980028927326202,\n -0.43828743696212770,\n -0.19539453089237213,\n 0.70773118734359740,\n -0.21142446994781494,\n -0.012780480086803436,\n 0.06678064167499542,\n -0.12673433125019073,\n -0.016046274453401566,\n 0.34845018386840820,\n -0.37029501795768740,\n -0.41383981704711914,\n 0.14042145013809204,\n -0.07714009284973145,\n 0.23623535037040710,\n -0.07902203500270844,\n -0.22669017314910890,\n 0.13354891538619995,\n 0.24691657721996307,\n -0.23718547821044922,\n 0.19054308533668518,\n -0.10418700426816940,\n 0.33725842833518980,\n -0.08150300383567810,\n -0.020559510216116905,\n -0.06275172531604767,\n 0.32859966158866880,\n -0.23827262222766876,\n 0.006859809160232544,\n -0.43626987934112550,\n 0.00422387570142746,\n -0.42823368310928345,\n 0.06333791464567184,\n 0.24973529577255250,\n -0.038866639137268066,\n 0.39266145229339600,\n -0.33822238445281980,\n 0.20358082652091980,\n 0.45638072490692140,\n 0.24203394353389740,\n -1.19550085067749020,\n -0.07111288607120514,\n 0.31889787316322327,\n -0.07209780812263489,\n 0.08574327081441879,\n -0.05155383050441742,\n 0.10396289825439453,\n 0.37258481979370117,\n 0.31418317556381226,\n -0.41241401433944700,\n 0.15978108346462250,\n 0.07924050092697144,\n 0.046578992158174515,\n 0.14554628729820251,\n 0.23571312427520752,\n -0.67628896236419680,\n 0.16695483028888702,\n -0.21981866657733917,\n 0.16223661601543427,\n -0.021025659516453743,\n 0.06956741958856583,\n -0.30338615179061890,\n -0.16539447009563446,\n 1.11133313179016110,\n 0.07875859737396240,\n 0.08710280805826187,\n 0.09269724786281586,\n 0.43146112561225890,\n -0.63181507587432860,\n 0.24583837389945984,\n -0.44661480188369750,\n 0.03440243750810623,\n 0.11665569245815277,\n 0.15819263458251953,\n 0.12243788689374924,\n 0.29396194219589233,\n 0.09694259613752365,\n 0.02281755954027176,\n -0.21637912094593048,\n -0.37317517399787903,\n 0.33211824297904970,\n 0.18465763330459595,\n 0.04658152908086777,\n -0.22277922928333282,\n -0.01132119633257389,\n -0.22123819589614868,\n 0.21734745800495148,\n 0.18564417958259583,\n -0.16151818633079530,\n -0.012778416275978088,\n 0.22715422511100770,\n -0.61123466491699220,\n -0.25164487957954407,\n -0.62368929386138920,\n -0.08976317942142487,\n -0.18425342440605164,\n 0.01974041759967804,\n 0.40116083621978760,\n 0.09709006547927856,\n 0.02330559492111206,\n -0.24683454632759094,\n 0.46549883484840393,\n 0.17286136746406555,\n 0.13242653012275696,\n -0.018801003694534302,\n 0.05591413378715515,\n -0.008567705750465393,\n 0.31677705049514770,\n -0.03249857574701309,\n 0.23144552111625670,\n -0.19878269731998444,\n 0.34026873111724854,\n 0.46859806776046753,\n 0.46318793296813965,\n -0.15173172950744630,\n 0.03247187286615372,\n -0.35306686162948610,\n 0.89938664436340330,\n -0.12891116738319397,\n 0.23192928731441498,\n -0.20485301315784454,\n 0.26682716608047485,\n 0.03473209589719772,\n -0.05685047805309296,\n 0.85817599296569820,\n 0.20728000998497010,\n -0.42015826702117920,\n -0.20325171947479248,\n -0.18902121484279633,\n -0.08646465092897415,\n 0.47624379396438600,\n 0.07658274471759796,\n 0.24890711903572083,\n 0.09770368039608002,\n -0.07584749162197113,\n -0.38329556584358215,\n -0.09186994284391403,\n 0.05287090688943863,\n -0.07528317719697952,\n -0.30612573027610780,\n 0.03865929692983627,\n 0.20936234295368195,\n -0.12896691262722015,\n 0.40101802349090576,\n -0.27413615584373474,\n -0.11920598894357681,\n -0.27652582526206970,\n 0.16087099909782410,\n 0.30647218227386475,\n -0.23265513777732850,\n -0.04036612808704376,\n -0.19249220192432404,\n 0.29576075077056885,\n -0.33208352327346800,\n -0.26545786857604980,\n -0.47843095660209656,\n 0.05080226808786392,\n -0.18463236093521118,\n -0.31534552574157715,\n -0.16215364634990692,\n 0.44137579202651980,\n 0.19857212901115417,\n -0.05095067620277405,\n 0.60800743103027340,\n 0.18676830828189850,\n 0.43479916453361510,\n -0.31042078137397766,\n 0.34246861934661865,\n -0.22135923802852630,\n -0.16289021074771880,\n 0.14682009816169740,\n -0.05069936811923981,\n -1.54986953735351560,\n -0.20729129016399384,\n 0.30774104595184326,\n -0.74974900484085080,\n 0.17779231071472168,\n 0.29844737052917480,\n -0.028225738555192947,\n 0.06633662432432175,\n 0.09315972030162811,\n 1.39255237579345700,\n 0.08268950879573822,\n 0.17041748762130737,\n -0.07805733382701874,\n -0.009891202673316002,\n -0.42962673306465150,\n -0.31951111555099490,\n -0.59171640872955320,\n 0.13881197571754456,\n 0.05963444709777832,\n 0.36311095952987670,\n 0.20484755933284760,\n -0.24004952609539032,\n -0.30473378300666810,\n -1.23349046707153320,\n 0.12950478494167328,\n 0.13113403320312500,\n -0.13356469571590424,\n -0.33658254146575930,\n -0.17962300777435303,\n 0.44429525732994080,\n 0.25605937838554380,\n -0.18255349993705750,\n -0.34041595458984375,\n 0.12862077355384827,\n -0.10708140581846237,\n -0.04143147915601730,\n -0.75912737846374510,\n 0.11328382045030594,\n -0.33319079875946045,\n -0.12773975729942322,\n -0.24314782023429870,\n -0.08418860286474228,\n 0.024868622422218323,\n -0.24916192889213562,\n 0.14967927336692810,\n 0.50531852245330810,\n -0.17624694108963013,\n -0.32372432947158813,\n 0.26036772131919860,\n -0.13107712566852570,\n -0.19526052474975586,\n -0.34996968507766724,\n 0.09007526934146881,\n -0.19009685516357422,\n -0.04918465018272400,\n 0.42598405480384827,\n -0.95772063732147220,\n -0.26188156008720400,\n -0.13462109863758087,\n -0.48766309022903440,\n -0.04416780546307564,\n -0.24330116808414460,\n 0.49982681870460510,\n -0.59761744737625120,\n -0.51605522632598880,\n 0.25707191228866577,\n 0.06983644515275955,\n -0.70166909694671630,\n -0.44174510240554810,\n 0.59883981943130490,\n -0.23916383087635040,\n -0.23846682906150818,\n -0.09755046665668488,\n 0.06088256835937500,\n 0.57148891687393190,\n -0.46913427114486694,\n 0.10756542533636093,\n 0.24516284465789795,\n -0.29232767224311830,\n -0.41785952448844910,\n -0.20276978611946106,\n -0.09826758503913880,\n 0.03998877853155136,\n 0.23493704199790955,\n -0.30350178480148315,\n -0.05809980630874634,\n -0.92815858125686650,\n -0.07727308571338654,\n 0.009639181196689606,\n 0.11947400122880936,\n -0.32069391012191770,\n -0.15333953499794006,\n -0.20000588893890380,\n 0.51839399337768550,\n -0.12886822223663330,\n -0.06542570143938065,\n -0.07890033721923828,\n 0.67580813169479370,\n 0.23756900429725647,\n -0.39474987983703613,\n -0.08904854953289032,\n 0.10808389633893967,\n 0.55502021312713620,\n -0.06922332942485810,\n 0.08530282974243164,\n -0.64823102951049800,\n -0.25969624519348145,\n 0.18102583289146423,\n -0.15270616114139557,\n -0.08661939203739166,\n -0.13988846540451050,\n -0.13920640945434570,\n -0.19092643260955810,\n -0.30402636528015137,\n 0.034376729279756546,\n 0.28963160514831543,\n -0.026797950267791748,\n -0.009581595659255981,\n -0.13647942245006560,\n 0.45053595304489136,\n -0.13274556398391724,\n -0.32779148221015930,\n -0.29205349087715150,\n -0.04352697730064392,\n -0.10386516153812408,\n 0.09681421518325806,\n -0.43473735451698303,\n -0.12714838981628418,\n -0.13591085374355316,\n -0.019401680678129196,\n 0.08398979157209396,\n -0.21437484025955200,\n 0.16760951280593872,\n 0.06209127604961395,\n 0.10363107919692993,\n 0.037978142499923706,\n 0.25537312030792236,\n -0.006652772426605225,\n -0.29459229111671450,\n 0.13102664053440094,\n 0.72616183757781980,\n -0.90198361873626710,\n -0.04998414218425751,\n 0.015783268958330154,\n 0.34867671132087710,\n ],\n \"clip_embeddings_magnitude\": 10.424424171447754,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[99, 135, 179]\",\n \"video_length\": None,\n \"removed\": False,\n \"in_trashcan\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"0\",\n \"size\": 9246412,\n \"subjectDistance\": 1.686,\n \"width\": 4032,\n \"main_file_id\": \"1f522608a263a88ea72010fe4f08f3071\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/e05dc030e807ae8fc31442fa2ba7fdf81.webp\",\n \"square_thumbnail\": \"square_thumbnails/e05dc030e807ae8fc31442fa2ba7fdf81.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/e05dc030e807ae8fc31442fa2ba7fdf81.webp\",\n \"added_on\": \"2023-06-16 16:30:42.859303 +00:00\",\n \"exif_gps_lat\": -33.8880555555556,\n \"exif_gps_lon\": 151.275180555556,\n \"exif_timestamp\": \"2017-10-18 15:08:09.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [151.275181, -33.888056],\n \"features\": [\n {\n \"id\": \"address.6965576203835628\",\n \"text\": \"Beach Road\",\n \"type\": \"Feature\",\n \"center\": [151.275216500055, -33.888012425],\n \"address\": \"17\",\n \"context\": [\n {\n \"id\": \"postcode.429582\",\n \"text\": \"2026\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCbzRP\",\n },\n {\n \"id\": \"locality.259156494\",\n \"text\": \"Bondi Beach\",\n \"wikidata\": \"Q673418\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpEM0pxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.275216500055, -33.888012425],\n },\n \"relevance\": 1,\n \"place_name\": \"17 Beach Road, Bondi Beach New South Wales 2026, Australia\",\n \"place_type\": [\"address\"],\n \"properties\": {\n \"accuracy\": \"rooftop\",\n \"mapbox_id\": \"dXJuOm1ieGFkcjo3NmQ1MTQzZi1kZmEwLTQ4NzAtYTAyNy0zYWY1MGJiZTIxYWU\",\n },\n },\n {\n \"id\": \"postcode.429582\",\n \"bbox\": [151.2582, -33.90058, 151.285977, -33.878047],\n \"text\": \"2026\",\n \"type\": \"Feature\",\n \"center\": [151.271075, -33.898745],\n \"context\": [\n {\n \"id\": \"locality.259156494\",\n \"text\": \"Bondi Beach\",\n \"wikidata\": \"Q673418\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpEM0pxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.271075, -33.898745],\n },\n \"relevance\": 1,\n \"place_name\": \"2026, Bondi Beach, New South Wales, Australia\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpCbzRP\"},\n },\n {\n \"id\": \"locality.259156494\",\n \"bbox\": [151.26203377, -33.896871333, 151.282858153, -33.884999032],\n \"text\": \"Bondi Beach\",\n \"type\": \"Feature\",\n \"center\": [151.2745, -33.8915],\n \"context\": [\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [151.2745, -33.8915]},\n \"relevance\": 1,\n \"place_name\": \"Bondi Beach, New South Wales, Australia\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q673418\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpEM0pxRGc\",\n },\n },\n {\n \"id\": \"place.24496142\",\n \"bbox\": [150.520934139, -34.11717528, 151.369884128, -33.562644328],\n \"text\": \"Sydney\",\n \"type\": \"Feature\",\n \"center\": [151.216454, -33.854816],\n \"context\": [\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.216454, -33.854816],\n },\n \"relevance\": 1,\n \"place_name\": \"Sydney, New South Wales, Australia\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n },\n {\n \"id\": \"region.33806\",\n \"bbox\": [140.999265, -37.5097258, 159.200456, -28.1370359],\n \"text\": \"New South Wales\",\n \"type\": \"Feature\",\n \"center\": [147.014694071448, -32.168971672412],\n \"context\": [\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [147.014694071448, -32.168971672412],\n },\n \"relevance\": 1,\n \"place_name\": \"New South Wales, Australia\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n },\n {\n \"id\": \"country.8718\",\n \"bbox\": [112.8256904, -54.8327658, 159.200456, -9.0436707],\n \"text\": \"Australia\",\n \"type\": \"Feature\",\n \"center\": [134.489562606981, -25.7349684916223],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [134.489562606981, -25.7349684916223],\n },\n \"relevance\": 1,\n \"place_name\": \"Australia\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Beach Road 2026 Bondi Beach Sydney New South Wales Australia\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"no horizon\",\n \"man made\",\n \"wood\",\n \"natural light\",\n \"cloth\",\n \"railing\",\n \"fencing\",\n \"soothing\",\n \"enclosed area\",\n ],\n \"categories\": [\"boardwalk\"],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"boardwalk , outdoor\",\n \"search_location\": \"Beach Road 2026 Bondi Beach Sydney New South Wales Australia\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n -0.23971192538738250,\n -0.06270129978656769,\n -0.26378244161605835,\n 0.06143847107887268,\n -0.17431882023811340,\n -0.45036801695823670,\n 0.23159477114677430,\n 0.45534604787826540,\n 0.41466632485389710,\n 0.43482506275177000,\n 0.033715225756168365,\n -0.12434051930904388,\n -0.02738797664642334,\n 0.13923697173595428,\n 0.71357935667037960,\n -0.10237488895654678,\n 0.44860103726387024,\n -0.12994986772537231,\n 0.16177919507026672,\n -0.52953779697418210,\n -0.15292952954769135,\n 0.35376715660095215,\n 0.68445837497711180,\n -0.12952817976474762,\n -0.65442270040512080,\n -0.11123260110616684,\n 0.35111707448959350,\n -0.21466320753097534,\n -0.07700178027153015,\n 0.60484236478805540,\n -0.35141080617904663,\n 0.42874458432197570,\n -0.14701411128044128,\n -0.31290709972381590,\n 0.41403374075889590,\n 0.42659783363342285,\n 0.14210951328277588,\n 0.21080046892166138,\n -0.20884071290493011,\n 1.55791330337524410,\n -1.02671170234680180,\n -0.21040523052215576,\n 0.21100938320159912,\n -0.018877588212490082,\n 0.045770272612571716,\n -1.74390268325805660,\n 0.27797201275825500,\n 0.11735321581363678,\n -0.16868110001087190,\n 0.07083947211503983,\n 0.17496472597122192,\n 0.18361502885818481,\n 0.06942721456289291,\n 0.29114612936973570,\n 0.16485445201396942,\n 0.35742586851119995,\n 0.71032893657684330,\n 0.08239606022834778,\n -0.48378357291221620,\n 0.40987241268157960,\n 0.61319535970687870,\n 0.07883536070585251,\n 0.52093547582626340,\n 0.56784075498580930,\n 0.13938774168491364,\n -0.15158170461654663,\n 0.37094724178314210,\n 0.83913630247116090,\n -0.48059833049774170,\n -0.05472422018647194,\n 0.50041580200195310,\n 0.14854013919830322,\n 0.11348950862884521,\n -0.10606382042169571,\n -0.032269977033138275,\n -0.10980830341577530,\n -0.06923694908618927,\n -0.44631308317184450,\n 0.06693774461746216,\n -0.67717289924621580,\n 0.16200356185436250,\n -0.08856121450662613,\n -0.45839691162109375,\n -0.59329897165298460,\n 0.027508363127708435,\n -0.17059496045112610,\n 1.00395536422729500,\n -0.40785533189773560,\n 0.20544975996017456,\n -0.61956644058227540,\n 0.30773437023162840,\n -0.75228130817413330,\n -6.36993885040283200,\n 0.23692467808723450,\n -0.18002653121948242,\n 0.41877511143684387,\n -0.03537013381719589,\n -0.13835334777832030,\n -0.28309312462806700,\n 0.99767374992370600,\n 0.43206804990768430,\n -0.61333036422729490,\n 0.31665223836898804,\n 0.011051952838897705,\n 0.26114824414253235,\n 0.20024497807025910,\n 0.86893093585968020,\n 0.16212932765483856,\n 0.038838282227516174,\n 0.10568191856145859,\n 0.18302081525325775,\n -0.49132484197616577,\n -0.13319587707519530,\n -0.38572445511817930,\n 0.09171506762504578,\n -0.26322168111801150,\n -0.66410481929779050,\n 0.13588872551918030,\n 0.47181224822998047,\n -0.05031928792595863,\n 0.25005054473876953,\n 0.21581515669822693,\n 0.25475072860717773,\n -0.24969539046287537,\n 0.54527783393859860,\n -0.27312040328979490,\n -0.55337727069854740,\n 0.23277127742767334,\n 0.014410212635993958,\n 0.24621416628360748,\n 0.11218813061714172,\n -0.35024920105934143,\n -0.21312740445137024,\n 0.80181252956390380,\n -0.79384481906890870,\n 0.50359219312667850,\n -0.45367711782455444,\n -0.48844304680824280,\n -0.11650219559669495,\n -0.12895251810550690,\n -0.26562160253524780,\n 0.20769277215003967,\n -0.21654091775417328,\n 0.21669554710388184,\n -0.23972092568874360,\n 0.43769764900207520,\n 0.32644972205162050,\n 0.35533404350280760,\n 0.28178536891937256,\n 0.40197637677192690,\n 0.16447728872299194,\n 0.41968211531639100,\n -1.00776052474975590,\n 0.09402647614479065,\n 0.07121045887470245,\n -0.38577982783317566,\n -0.15076629817485810,\n -0.67609924077987670,\n -0.37826904654502870,\n -0.17423087358474731,\n -0.019086994230747223,\n 0.09864071011543274,\n 0.30087676644325256,\n 0.27617043256759644,\n 0.06742041558027267,\n -0.01719839498400688,\n 0.47258347272872925,\n 0.44886517524719240,\n 0.06894390285015106,\n 0.013095356523990631,\n -0.34940755367279050,\n 0.02119274064898491,\n 0.44170263409614563,\n -0.17398183047771454,\n 0.11397088319063187,\n -0.25963869690895080,\n 0.66773271560668950,\n -0.029286637902259827,\n 0.26047524809837340,\n 0.16783103346824646,\n -0.10030107200145721,\n -0.28766095638275146,\n -0.20763801038265228,\n -0.020237110555171967,\n 0.26628720760345460,\n -0.03873760998249054,\n 0.33926361799240110,\n -0.22293141484260560,\n -0.02787308767437935,\n -0.02593088522553444,\n -0.14129941165447235,\n 0.26731464266777040,\n -0.08877785503864288,\n 0.31370738148689270,\n -0.03215152025222778,\n 0.28753554821014404,\n 0.28241288661956787,\n -0.58873462677001950,\n -1.19819414615631100,\n -0.31043520569801330,\n 0.52656930685043330,\n -0.15177714824676514,\n 0.04967714473605156,\n 0.21121788024902344,\n -0.34197264909744260,\n -0.58301079273223880,\n -0.05980183556675911,\n 0.08901736885309220,\n -0.016958534717559814,\n 0.70352792739868160,\n 0.04605954885482788,\n 0.33565390110015870,\n -0.15496665239334106,\n 0.008393734693527222,\n -0.39955216646194460,\n -0.58763426542282100,\n -0.06084361672401428,\n 0.49945196509361267,\n 1.24120748043060300,\n -0.26017981767654420,\n 0.88132452964782710,\n 0.42076665163040160,\n -0.11274620890617370,\n -0.17265769839286804,\n -0.42378520965576170,\n 0.51237106323242190,\n -0.05410161614418030,\n 0.19328644871711730,\n 0.06429362297058105,\n 0.08922037482261658,\n 0.26436531543731690,\n -0.07118786871433258,\n 0.16370669007301330,\n 0.53644728660583500,\n -0.22373552620410920,\n -0.10707579553127289,\n -0.40309336781501770,\n -0.36508309841156006,\n -0.35035932064056396,\n -0.40398806333541870,\n -0.19331835210323334,\n 0.46320116519927980,\n 0.38004130125045776,\n 0.28961494565010070,\n -0.11237305402755737,\n 0.92902261018753050,\n 0.52717500925064090,\n -0.55640512704849240,\n 0.43815851211547850,\n 0.17259076237678528,\n 0.16953691840171814,\n -0.15717472136020660,\n 0.36117780208587646,\n -0.15650644898414612,\n 0.41624057292938230,\n 0.23601931333541870,\n -0.07374796271324158,\n 0.19807760417461395,\n 0.90526473522186280,\n 0.15534612536430360,\n -0.10369679331779480,\n 0.44986745715141296,\n 0.14716669917106628,\n -1.63226437568664550,\n 0.34298181533813477,\n -0.07653657346963882,\n 0.09407752752304077,\n -0.09107702970504761,\n -0.15198035538196564,\n 0.40160834789276123,\n -0.11560469865798950,\n 0.12189065665006638,\n 0.55020934343338010,\n 0.23245319724082947,\n -0.33362627029418945,\n -0.026688329875469208,\n 0.036947548389434814,\n -0.15187057852745056,\n 0.16617432236671448,\n -0.22867006063461304,\n -0.14950640499591827,\n 0.18319699168205260,\n -0.16728959977626800,\n -0.10700336098670960,\n -0.40600830316543580,\n 0.11292800307273865,\n -0.45587986707687380,\n 0.08800569921731949,\n -0.12228171527385712,\n -0.20845821499824524,\n 0.012005604803562164,\n 0.009173348546028137,\n 0.03591999411582947,\n 0.06969346106052399,\n -0.057597748935222626,\n -0.26176500320434570,\n 0.057852864265441895,\n 0.52407950162887570,\n -0.06004469096660614,\n 0.09566502273082733,\n 0.45640081167221070,\n -0.36844995617866516,\n -0.19065113365650177,\n -0.24948999285697937,\n 0.10381768643856049,\n 0.20994564890861510,\n -0.43057119846343994,\n 0.44703856110572815,\n 0.45062947273254395,\n -0.048142626881599426,\n -0.43613320589065550,\n 0.68679362535476680,\n 0.80048835277557370,\n -0.27635425329208374,\n 0.05641265958547592,\n 0.45497626066207886,\n -0.07284688949584961,\n 0.027119621634483337,\n -0.71727848052978520,\n 0.051672548055648804,\n 0.49660560488700867,\n -0.014366000890731812,\n -0.32767134904861450,\n 0.57496893405914310,\n -0.10740306973457336,\n 0.31921407580375670,\n -0.18982721865177155,\n 0.62163728475570680,\n 0.05321182310581207,\n -0.007055327296257019,\n -0.39857268333435060,\n -0.42034840583801270,\n 0.42702692747116090,\n -0.25245079398155210,\n -0.15837675333023070,\n 0.13288237154483795,\n 0.38176494836807250,\n -0.002602334599941969,\n -0.13858896493911743,\n -0.14321827888488770,\n -0.22877745330333710,\n -0.49555492401123047,\n 0.05159477889537811,\n 0.11018384248018265,\n 0.51627355813980100,\n 0.27100521326065063,\n 0.25322765111923220,\n 0.24974197149276733,\n 0.33265298604965210,\n -0.29485535621643066,\n -1.29044723510742190,\n 0.04591430723667145,\n -0.07321572303771973,\n -0.14710950851440430,\n 0.12631073594093323,\n 1.13345527648925780,\n -0.014754462987184525,\n -0.29979729652404785,\n 0.16985306143760680,\n -0.009212018921971321,\n 0.07431145012378693,\n -0.09315565973520279,\n 0.009219720959663391,\n 0.010067738592624664,\n -0.63250660896301270,\n 0.17347553372383118,\n -0.07402996718883514,\n -0.71398895978927610,\n -0.21345643699169160,\n -0.05977027118206024,\n -0.15027517080307007,\n 0.23419082164764404,\n 0.07772324234247208,\n 0.24337679147720337,\n -0.21138656139373780,\n -0.14810818433761597,\n 0.86943644285202030,\n 0.16901102662086487,\n -0.56860476732254030,\n 0.46403664350509644,\n 0.09731522202491760,\n -0.56933224201202390,\n -0.12460615485906601,\n -0.11898320913314820,\n 0.07008326053619385,\n -0.28138583898544310,\n -0.13215136528015137,\n -0.17825207114219666,\n 0.42181590199470520,\n -1.26650285720825200,\n -0.38550242781639100,\n -0.79046458005905150,\n 0.27815857529640200,\n -0.33496844768524170,\n 0.35100549459457400,\n -0.047406621277332306,\n -0.28952226042747500,\n 0.05926326662302017,\n -0.40099292993545530,\n -0.67989456653594970,\n -0.10077974200248718,\n 0.20837950706481934,\n 0.22934542596340180,\n -0.19858762621879578,\n 0.03676336631178856,\n -0.35527148842811584,\n 0.16404820978641510,\n 0.00022447854280471802,\n 0.76411741971969600,\n -0.28233665227890015,\n 0.63915759325027470,\n -0.37327551841735840,\n -0.16160306334495544,\n 0.20690485835075378,\n 0.13408637046813965,\n 0.04813005030155182,\n -0.20736610889434814,\n -0.04153786227107048,\n 0.45611751079559326,\n -0.16623178124427795,\n 0.62960529327392580,\n 0.05700150877237320,\n 0.31719794869422910,\n -0.85579967498779300,\n 0.26484364271163940,\n -0.24446523189544678,\n -0.15408588945865630,\n -0.19397708773612976,\n -0.11019599437713623,\n 0.03778602182865143,\n -0.29206633567810060,\n 0.01961149275302887,\n 0.15172131359577180,\n -0.28065830469131470,\n 0.11242515593767166,\n -0.18694521486759186,\n -0.58477157354354860,\n 0.05389549210667610,\n 0.22777841985225677,\n -0.32870626449584960,\n 0.29864984750747680,\n 0.18124125897884370,\n -0.20861062407493590,\n 0.0054992809891700745,\n 0.03267093002796173,\n -0.49986141920089720,\n -0.11108686029911041,\n -0.59358328580856320,\n -0.37555414438247680,\n -0.06972122937440872,\n 0.20261085033416748,\n 0.14063805341720580,\n -0.03213777393102646,\n 0.64066505432128910,\n 0.12838377058506012,\n -0.26487171649932860,\n 0.08924877643585205,\n 0.24947234988212585,\n 0.15273018181324005,\n 0.40036311745643616,\n 0.044563665986061096,\n 0.06986735761165619,\n -0.06490179151296616,\n -0.22339831292629242,\n 0.84133744239807130,\n 0.18982148170471191,\n -0.25457212328910830,\n 0.005809254944324493,\n -0.10496775060892105,\n 0.09663195163011551,\n 0.053644873201847076,\n -0.12464396655559540,\n -0.52167260646820070,\n -0.03012670949101448,\n -0.42671746015548706,\n 0.04330782592296600,\n -0.15239623188972473,\n 0.24043066799640656,\n -0.27606680989265440,\n -0.21640032529830933,\n 0.22917535901069640,\n -0.07214342057704926,\n -0.007351040840148926,\n -0.10876821726560593,\n -0.30059793591499330,\n 0.17446355521678925,\n -0.11450521647930145,\n -0.21727171540260315,\n 0.13252682983875275,\n 0.28893244266510010,\n -0.34937959909439087,\n 0.10954467207193375,\n 0.67957222461700440,\n -0.40904277563095090,\n -0.005858458578586578,\n -0.17941795289516450,\n -0.02339877001941204,\n -0.007561013102531433,\n -0.34222379326820374,\n -0.23668894171714783,\n -0.19201411306858063,\n 0.07832273840904236,\n -0.17331452667713165,\n 0.15513053536415100,\n 0.20789094269275665,\n 0.011405497789382935,\n -0.022110439836978912,\n 0.31066673994064330,\n -0.89273244142532350,\n 0.69325125217437740,\n 0.22300001978874207,\n -0.06441357731819153,\n ],\n \"clip_embeddings_magnitude\": 10.693284034729004,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[64, 59, 54]\",\n \"video_length\": None,\n \"removed\": False,\n \"in_trashcan\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"1/390\",\n \"size\": 9145064,\n \"subjectDistance\": 0.377,\n \"width\": 4032,\n \"main_file_id\": \"e05dc030e807ae8fc31442fa2ba7fdf81\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/c57f08a55c6daa689e6e13a1bf8f81471.webp\",\n \"square_thumbnail\": \"square_thumbnails/c57f08a55c6daa689e6e13a1bf8f81471.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/c57f08a55c6daa689e6e13a1bf8f81471.webp\",\n \"added_on\": \"2023-06-16 16:30:49.679482 +00:00\",\n \"exif_gps_lat\": 33.91345,\n \"exif_gps_lon\": 78.4573888888889,\n \"exif_timestamp\": \"2017-10-12 08:00:51.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [78.457389, 33.91345],\n \"features\": [\n {\n \"id\": \"address.4021415981422040\",\n \"text\": \"Lakeshore Road\",\n \"type\": \"Feature\",\n \"center\": [78.45710764679163, 33.91336849412403],\n \"context\": [\n {\n \"id\": \"postcode.13119083\",\n \"text\": \"194201\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzp5QzVy\",\n },\n {\n \"id\": \"locality.3711994475\",\n \"text\": \"Shachokol\",\n \"wikidata\": \"Q24912826\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzozVUNLYXc\",\n },\n {\n \"id\": \"place.25135211\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [78.45710764679163, 33.91336849412403],\n },\n \"relevance\": 1,\n \"place_name\": \"Lakeshore Road ، 194201 Leh، India\",\n \"place_type\": [\"address\"],\n \"properties\": {\"accuracy\": \"street\"},\n },\n {\n \"id\": \"postcode.13119083\",\n \"bbox\": [77.159628, 32.619815, 79.042702, 34.608467],\n \"text\": \"194201\",\n \"type\": \"Feature\",\n \"center\": [77.668644, 34.050466],\n \"context\": [\n {\n \"id\": \"locality.3711994475\",\n \"text\": \"Shachokol\",\n \"wikidata\": \"Q24912826\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzozVUNLYXc\",\n },\n {\n \"id\": \"place.25135211\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.668644, 34.050466],\n },\n \"relevance\": 1,\n \"place_name\": \"194201, Leh, Ladakh, India\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzp5QzVy\"},\n },\n {\n \"id\": \"locality.3711994475\",\n \"bbox\": [77.324409401, 33.676681361, 79.042702, 35.496086],\n \"text\": \"Shachokol\",\n \"type\": \"Feature\",\n \"center\": [78.167596, 34.039239],\n \"context\": [\n {\n \"id\": \"place.25135211\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [78.167596, 34.039239],\n },\n \"relevance\": 1,\n \"place_name\": \"Shachokol, Leh, Leh, Ladakh, India\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q24912826\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzozVUNLYXc\",\n },\n },\n {\n \"id\": \"place.25135211\",\n \"bbox\": [77.107033, 32.33574, 79.305839, 35.522256],\n \"text\": \"Leh\",\n \"type\": \"Feature\",\n \"center\": [77.584813, 34.164203],\n \"context\": [\n {\n \"id\": \"district.3196523\",\n \"text\": \"Leh\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n },\n {\n \"id\": \"region.222315\",\n \"text\": \"Ladakh\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.584813, 34.164203],\n },\n \"relevance\": 1,\n \"place_name\": \"Leh, Ladakh, India\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n },\n },\n {\n \"id\": \"region.222315\",\n \"bbox\": [75.306629, 32.33574, 79.305851, 35.673315],\n \"text\": \"Ladakh\",\n \"type\": \"Feature\",\n \"center\": [77.27783203125, 34.0071350643588],\n \"context\": [\n {\n \"id\": \"country.8811\",\n \"text\": \"India\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [77.27783203125, 34.0071350643588],\n },\n \"relevance\": 1,\n \"place_name\": \"Ladakh, India\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"short_code\": \"IN-LA\",\n },\n },\n {\n \"id\": \"country.8811\",\n \"bbox\": [68.1152344, 6.6718373, 97.395359, 35.673315],\n \"text\": \"India\",\n \"type\": \"Feature\",\n \"center\": [78.476681027237, 22.1991660760527],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [78.476681027237, 22.1991660760527],\n },\n \"relevance\": 1,\n \"place_name\": \"India\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"short_code\": \"in\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Lakeshore Road 194201 Shachokol Leh Ladakh India\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"man made\",\n \"natural light\",\n \"open area\",\n \"no horizon\",\n \"clouds\",\n \"cloth\",\n \"vertical components\",\n \"sunny\",\n \"plastic\",\n ],\n \"categories\": [\"playground\"],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"playground , outdoor\",\n \"search_location\": \"Lakeshore Road 194201 Shachokol Leh Ladakh India\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n -0.18379329144954681,\n 0.14253073930740356,\n -0.34837892651557920,\n 0.37873998284339905,\n 0.34088349342346190,\n 0.17458681762218475,\n 0.25472477078437805,\n 0.29220539331436160,\n -0.40241336822509766,\n -0.11195755004882812,\n 0.30859681963920593,\n 0.04621639847755432,\n 0.15303245186805725,\n -0.42163884639739990,\n -0.18283540010452270,\n 0.27533942461013794,\n -1.12763762474060060,\n -0.24703232944011688,\n 0.023488599807024002,\n -0.51369655132293700,\n -0.18228112161159515,\n 0.04996514320373535,\n 0.09342576563358307,\n -0.17483948171138763,\n -0.25493249297142030,\n 0.019286181777715683,\n 0.17102202773094177,\n -0.29159709811210630,\n 0.33772629499435425,\n -0.19569182395935059,\n -0.19514861702919006,\n 0.35274094343185425,\n -0.05648523569107056,\n -0.30300104618072510,\n 0.01596139743924141,\n 0.47075837850570680,\n -0.26226904988288880,\n 0.10937485098838806,\n -0.45466923713684080,\n -0.61406242847442630,\n 0.045746058225631714,\n -0.09549369663000107,\n 0.49270814657211304,\n 0.02534257248044014,\n -0.29359999299049380,\n -0.07753305882215500,\n -0.02130131423473358,\n -0.07495892047882080,\n -0.60064113140106200,\n -0.014182962477207184,\n 0.18451625108718872,\n 0.35400345921516420,\n 0.22977495193481445,\n 0.06056281179189682,\n -0.16951103508472443,\n 0.43877935409545900,\n -0.08079877495765686,\n -0.07902894914150238,\n -0.05741514265537262,\n 0.55906730890274050,\n 0.58910524845123290,\n -0.26552647352218630,\n 0.47798049449920654,\n -0.17932428419589996,\n 0.03466450423002243,\n -0.32181957364082336,\n 0.32691439986228943,\n 1.57434558868408200,\n -0.60944569110870360,\n -0.30789577960968020,\n -0.03525789454579353,\n 0.21574194729328156,\n 0.09106104075908661,\n 0.07360906153917313,\n -0.21098615229129790,\n 0.14903579652309418,\n -0.20697991549968720,\n 0.09246206283569336,\n -0.30049309134483340,\n -0.25683689117431640,\n -0.42003977298736570,\n 0.38077199459075930,\n -0.32969945669174194,\n -0.46575132012367250,\n 0.24943169951438904,\n 0.38942575454711914,\n -0.37571138143539430,\n -0.58096361160278320,\n 0.11415434628725052,\n -0.42314538359642030,\n 0.30862617492675780,\n 0.26128217577934265,\n -7.43450069427490200,\n -0.14982101321220398,\n -0.007992172613739967,\n 0.12065914273262024,\n -0.38246965408325195,\n -0.34107217192649840,\n -0.80451864004135130,\n -1.04886448383331300,\n 0.55530327558517460,\n -0.24028851091861725,\n 0.14857441186904907,\n 0.07519923150539398,\n -0.44258895516395570,\n -0.22696885466575623,\n -1.76246547698974600,\n -0.04149024933576584,\n -0.33089119195938110,\n 0.034649670124053955,\n 0.20300607383251190,\n -0.18623992800712585,\n 0.18344642221927643,\n 0.23681724071502686,\n 0.27303919196128845,\n 0.08612959086894989,\n 0.08668576925992966,\n 0.09101299196481705,\n 0.43840888142585754,\n -0.61762166023254400,\n -0.20999190211296082,\n 0.46851450204849243,\n -0.060045816004276276,\n -0.10672970116138458,\n -0.12689085304737090,\n -0.22409234941005707,\n -0.28366869688034060,\n -0.25668996572494507,\n 0.04891391843557358,\n 0.35491180419921875,\n 0.27223268151283264,\n -0.19807010889053345,\n 0.01612691581249237,\n 0.98669499158859250,\n 0.94963777065277100,\n 0.11202965676784515,\n 0.13241872191429138,\n -0.33038973808288574,\n -0.24429757893085480,\n 0.30757093429565430,\n -0.10918612778186798,\n -0.20302526652812958,\n -0.014930151402950287,\n 0.67508393526077270,\n -0.30977970361709595,\n -0.19329161942005157,\n 0.04983800649642944,\n 0.58730208873748780,\n -0.76442992687225340,\n 0.10625885426998138,\n -0.009232066571712494,\n 0.26251035928726196,\n 0.29365420341491700,\n -0.28812190890312195,\n 0.07605545222759247,\n -0.65482407808303830,\n 0.35084849596023560,\n 0.34299698472023010,\n 0.33285912871360780,\n -0.22953245043754578,\n -0.20801588892936707,\n -0.05146145820617676,\n 0.18107087910175323,\n 0.51615625619888310,\n -0.28014412522315980,\n 0.21909615397453308,\n 1.08348631858825680,\n 0.43758583068847656,\n -0.15551409125328064,\n -0.11162629723548889,\n 0.08055899292230606,\n 0.09409813582897186,\n 0.021383360028266907,\n -0.50918841361999510,\n -0.11208890378475189,\n -0.38599658012390137,\n -0.86290252208709720,\n 0.24611333012580872,\n -0.62409228086471560,\n 0.27000692486763000,\n -0.019148532301187515,\n -0.15995115041732788,\n 0.0021793851628899574,\n 0.07647107541561127,\n 0.19650121033191680,\n 0.25712081789970400,\n 0.15782865881919860,\n -0.12753826379776000,\n 0.13912162184715270,\n 0.057879120111465454,\n 0.12957704067230225,\n 0.23146329820156097,\n 0.024382056668400764,\n 0.04920619726181030,\n -0.24734464287757874,\n 0.23542645573616028,\n -0.17139248549938202,\n -0.16124279797077180,\n 0.63887059688568120,\n 0.39621770381927490,\n -0.20532843470573425,\n 0.012457119300961494,\n -0.39233744144439700,\n 0.71717041730880740,\n 0.25644633173942566,\n -0.49303287267684937,\n -0.04727388173341751,\n -0.38488596677780150,\n 0.03999908268451691,\n -0.11684153228998184,\n -1.22885262966156000,\n -0.15317772328853607,\n 0.03499954938888550,\n 0.18243020772933960,\n -0.32903003692626953,\n -0.28470265865325930,\n -0.17769415676593780,\n 0.35230401158332825,\n -0.68235456943511960,\n -0.44540110230445860,\n 0.25034108757972720,\n 0.23645302653312683,\n -0.09606185555458069,\n 0.23800262808799744,\n 0.11027810722589493,\n 0.05310886353254318,\n -0.15807375311851501,\n -0.32893413305282590,\n -0.12554939091205597,\n -0.09278482198715210,\n 0.36654987931251526,\n 0.31478390097618103,\n 0.022500615566968918,\n -0.17153693735599518,\n 0.58437371253967290,\n -0.56753563880920410,\n 0.0016302019357681274,\n -0.31919938325881960,\n -0.01767304539680481,\n -0.30621975660324097,\n 0.42181178927421570,\n 0.18961273133754730,\n -0.27777704596519470,\n -0.04874429106712341,\n -0.83623814582824710,\n -0.59505325555801390,\n 0.15763781964778900,\n 0.11460117995738983,\n -0.037775248289108276,\n 0.36981520056724550,\n 0.04750179499387741,\n 0.28530880808830260,\n -0.32593557238578796,\n 0.20472106337547302,\n -0.14085072278976440,\n -0.43984699249267580,\n -0.44318401813507080,\n 0.16889306902885437,\n 0.69945454597473140,\n 0.022213801741600037,\n 0.04395437240600586,\n 0.018100500106811523,\n -0.60659039020538330,\n 0.65107047557830810,\n 0.23823469877243042,\n -0.17507782578468323,\n 0.08020025491714478,\n 0.31494301557540894,\n 0.30044376850128174,\n -0.23546507954597473,\n 0.11599977314472198,\n 0.48960816860198975,\n 0.23746117949485780,\n -0.48635044693946840,\n -0.23878610134124756,\n 0.26654928922653200,\n -0.25886836647987366,\n -0.08302380144596100,\n -0.13755083084106445,\n 0.24222144484519958,\n 0.47945415973663330,\n 0.38373395800590515,\n 0.30907928943634033,\n -0.25348061323165894,\n 0.12096904218196869,\n 0.28086936473846436,\n -1.17676830291748050,\n 0.04671703651547432,\n -0.48752713203430176,\n -0.20467823743820190,\n 0.37203326821327210,\n -0.58486735820770260,\n 0.17445452511310577,\n -0.08156547695398330,\n 0.45127516984939575,\n -0.02351894974708557,\n -0.19682705402374268,\n 0.17286786437034607,\n -0.17587804794311523,\n 0.30918318033218384,\n 0.15095241367816925,\n -0.22093287110328674,\n -0.08669745922088623,\n -0.11007970571517944,\n 0.003908701241016388,\n -0.44974136352539060,\n 0.13072070479393005,\n 0.77382683753967290,\n 0.30658128857612610,\n -0.24635344743728638,\n -0.21027483046054840,\n 0.16422425210475922,\n 0.98671263456344600,\n -0.42050862312316895,\n 0.22445756196975708,\n 0.23323085904121400,\n 0.21030767261981964,\n 0.08243256807327270,\n 0.18058310449123383,\n -0.55228275060653690,\n 0.029390739277005196,\n -0.0027499347925186157,\n -0.26178988814353943,\n 0.39981740713119507,\n -0.46446281671524050,\n 0.20612671971321106,\n 0.02648659050464630,\n -0.06571706384420395,\n -0.09742008894681930,\n 0.27914172410964966,\n -0.44574266672134400,\n -0.004618685692548752,\n -0.20941592752933502,\n -0.014536745846271515,\n 0.04214660823345184,\n 0.16787378489971160,\n 0.12518627941608430,\n 0.14714732766151428,\n 0.42325642704963684,\n -0.18695265054702760,\n 0.59449142217636110,\n -0.41381680965423584,\n -0.08633117377758026,\n 0.48078146576881410,\n 0.01567218452692032,\n -0.11084793508052826,\n -0.25879496335983276,\n 0.25035518407821655,\n 0.08629626780748367,\n -0.17640779912471770,\n 0.35798883438110350,\n 0.003633505664765835,\n 0.26752072572708130,\n -0.008670955896377563,\n -0.060069844126701355,\n -0.45262604951858520,\n 0.0029344558715820312,\n 0.81401431560516360,\n 0.30656427145004270,\n 0.10814795643091202,\n 0.16015036404132843,\n 0.27728241682052610,\n 0.003430556505918503,\n 0.59562069177627560,\n -0.60939180850982670,\n 0.11420226097106934,\n -0.07706215977668762,\n -0.68500268459320070,\n -0.005410104990005493,\n 0.013373695313930511,\n 0.21606793999671936,\n 0.08226697891950607,\n -0.19103342294692993,\n 0.37374818325042725,\n 0.44795745611190796,\n -0.11685820668935776,\n 1.60860514640808100,\n -0.28285527229309080,\n 0.58519774675369260,\n 0.17927187681198120,\n 0.43113380670547485,\n -0.08384191989898682,\n -0.23026300966739655,\n -0.87972706556320190,\n -0.0010712891817092896,\n -0.11425804346799850,\n -0.03720918670296669,\n -0.013472747057676315,\n -0.20737977325916290,\n 1.80557346343994140,\n -0.64729046821594240,\n -0.30271497368812560,\n -0.061230555176734924,\n -0.38844275474548340,\n 0.17786000669002533,\n -0.031046316027641296,\n 0.24852822721004486,\n -0.39031293988227844,\n -0.28145694732666016,\n -0.52712631225585940,\n 0.43997967243194580,\n -0.22887703776359558,\n 0.17240211367607117,\n -0.29442641139030457,\n 0.35376852750778200,\n 0.22806043922901154,\n -0.06477281451225281,\n -0.35966825485229490,\n 0.40587043762207030,\n -0.08254503458738327,\n 0.93505430221557620,\n 0.013737834990024567,\n -0.16866286098957062,\n -0.17242559790611267,\n -0.09762792289257050,\n 0.06496423482894897,\n -0.53277724981307980,\n -0.41173180937767030,\n -0.10729959607124329,\n -0.22943912446498870,\n 0.09855599701404572,\n -0.31594187021255493,\n 0.63642692565917970,\n 0.08816694468259811,\n -0.11014832556247711,\n -0.28573364019393920,\n -0.16491752862930298,\n -0.013046562671661377,\n -0.57793509960174560,\n 0.16094970703125000,\n -0.41850280761718750,\n -0.07970791310071945,\n 0.15436783432960510,\n -0.29503697156906130,\n -0.052088260650634766,\n -0.33966383337974550,\n 0.018201902508735657,\n -0.06606295704841614,\n 0.04100047051906586,\n 0.36471304297447205,\n -0.40008947253227234,\n 0.07975389063358307,\n 0.39990115165710450,\n 0.15097494423389435,\n -0.21128392219543457,\n 0.05080147460103035,\n -0.046003326773643494,\n 0.25351849198341370,\n 0.16474604606628418,\n 0.08277872204780579,\n 0.22466805577278137,\n 0.22473768889904022,\n -0.13952386379241943,\n -0.35128515958786010,\n -0.17046678066253662,\n -0.04700194299221039,\n 0.21633064746856690,\n -0.27318200469017030,\n 0.02723865583539009,\n -0.47086358070373535,\n 0.25049856305122375,\n -0.48574218153953550,\n -0.08048862218856812,\n -0.36891525983810425,\n 0.03864942863583565,\n -0.42741003632545470,\n 0.74072462320327760,\n -0.010117409750819206,\n 0.19975064694881440,\n 0.03020879626274109,\n 0.06506866216659546,\n -0.04526616632938385,\n 0.26060470938682556,\n -0.43580517172813416,\n -0.04713132977485657,\n -0.33256918191909790,\n -0.17320886254310608,\n -0.22390314936637878,\n -0.31866759061813354,\n -0.29139828681945800,\n 0.07238420099020004,\n -0.07740981876850128,\n 0.11880752444267273,\n -0.26510918140411377,\n 0.15372966229915620,\n 0.41496598720550537,\n -0.18698418140411377,\n 0.0033263303339481354,\n -0.50970172882080080,\n 0.15158197283744812,\n -0.06647455692291260,\n 0.21226441860198975,\n 0.47759312391281130,\n -0.82750320434570310,\n -0.008479125797748566,\n -0.64633381366729740,\n -0.16696789860725403,\n 0.012448564171791077,\n -0.03979574143886566,\n -0.029009684920310974,\n -0.84793686866760250,\n 0.27539581060409546,\n -0.03149215131998062,\n 0.12921379506587982,\n 0.14929755032062530,\n -0.19812837243080140,\n -0.20390477776527405,\n 0.030443966388702393,\n -0.46114227175712585,\n 0.15821106731891632,\n -0.05045306682586670,\n -0.057202182710170746,\n ],\n \"clip_embeddings_magnitude\": 11.107995986938477,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[126, 149, 181]\",\n \"video_length\": None,\n \"removed\": False,\n \"in_trashcan\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"0\",\n \"size\": 6666687,\n \"subjectDistance\": 4.691,\n \"width\": 4032,\n \"main_file_id\": \"c57f08a55c6daa689e6e13a1bf8f81471\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/c4d896c9c9c0ca3c312b19f2f22d69ae1.webp\",\n \"square_thumbnail\": \"square_thumbnails/c4d896c9c9c0ca3c312b19f2f22d69ae1.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/c4d896c9c9c0ca3c312b19f2f22d69ae1.webp\",\n \"added_on\": \"2023-06-16 16:30:53.937783 +00:00\",\n \"exif_gps_lat\": 44.5542,\n \"exif_gps_lon\": -78.1953472222222,\n \"exif_timestamp\": \"2017-08-17 17:59:35.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [-78.195347, 44.5542],\n \"features\": [\n {\n \"id\": \"address.2311178160127006\",\n \"text\": \"Fire Route 47\",\n \"type\": \"Feature\",\n \"center\": [-78.1955566, 44.5540639],\n \"address\": \"2880\",\n \"context\": [\n {\n \"id\": \"postcode.2669825575\",\n \"text\": \"K0L 2H0\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpueUpPSnc\",\n },\n {\n \"id\": \"place.106145831\",\n \"text\": \"Lakefield\",\n \"wikidata\": \"Q114506323\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCbE9vSnc\",\n },\n {\n \"id\": \"district.1140263\",\n \"text\": \"Peterborough County\",\n \"wikidata\": \"Q730542\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFV1lu\",\n },\n {\n \"id\": \"region.17447\",\n \"text\": \"Ontario\",\n \"wikidata\": \"Q1904\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpSQ2M\",\n \"short_code\": \"CA-ON\",\n },\n {\n \"id\": \"country.8743\",\n \"text\": \"Canada\",\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"short_code\": \"ca\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [-78.1955566, 44.5540639],\n },\n \"relevance\": 1,\n \"place_name\": \"2880 Fire Route 47, Lakefield, Ontario K0L 2H0, Canada\",\n \"place_type\": [\"address\"],\n \"properties\": {\n \"accuracy\": \"point\",\n \"mapbox_id\": \"dXJuOm1ieGFkcjoyOTlkZDc4Yy0wYWVlLTQ2ZGUtYmFkOC04ZDA0Y2VhYzNkZGQ\",\n },\n },\n {\n \"id\": \"postcode.2669825575\",\n \"bbox\": [-78.383699659, 44.336123836, -77.917196264, 44.674985335],\n \"text\": \"K0L 2H0\",\n \"type\": \"Feature\",\n \"center\": [-78.27, 44.42],\n \"context\": [\n {\n \"id\": \"place.106145831\",\n \"text\": \"Lakefield\",\n \"wikidata\": \"Q114506323\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCbE9vSnc\",\n },\n {\n \"id\": \"district.1140263\",\n \"text\": \"Peterborough County\",\n \"wikidata\": \"Q730542\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFV1lu\",\n },\n {\n \"id\": \"region.17447\",\n \"text\": \"Ontario\",\n \"wikidata\": \"Q1904\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpSQ2M\",\n \"short_code\": \"CA-ON\",\n },\n {\n \"id\": \"country.8743\",\n \"text\": \"Canada\",\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"short_code\": \"ca\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [-78.27, 44.42]},\n \"relevance\": 1,\n \"place_name\": \"K0L 2H0, Lakefield, Ontario, Canada\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpueUpPSnc\"},\n },\n {\n \"id\": \"place.106145831\",\n \"bbox\": [-78.383699659, 44.402393807, -78.086446707, 44.59913282],\n \"text\": \"Lakefield\",\n \"type\": \"Feature\",\n \"center\": [-78.272195, 44.423337],\n \"context\": [\n {\n \"id\": \"district.1140263\",\n \"text\": \"Peterborough County\",\n \"wikidata\": \"Q730542\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFV1lu\",\n },\n {\n \"id\": \"region.17447\",\n \"text\": \"Ontario\",\n \"wikidata\": \"Q1904\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpSQ2M\",\n \"short_code\": \"CA-ON\",\n },\n {\n \"id\": \"country.8743\",\n \"text\": \"Canada\",\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"short_code\": \"ca\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [-78.272195, 44.423337],\n },\n \"relevance\": 1,\n \"place_name\": \"Lakefield, Ontario, Canada\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q114506323\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCbE9vSnc\",\n },\n },\n {\n \"id\": \"district.1140263\",\n \"bbox\": [-78.654828, 44.081278, -77.727372, 44.916769],\n \"text\": \"Peterborough County\",\n \"type\": \"Feature\",\n \"center\": [-78.332479, 44.302625],\n \"context\": [\n {\n \"id\": \"region.17447\",\n \"text\": \"Ontario\",\n \"wikidata\": \"Q1904\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpSQ2M\",\n \"short_code\": \"CA-ON\",\n },\n {\n \"id\": \"country.8743\",\n \"text\": \"Canada\",\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"short_code\": \"ca\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [-78.332479, 44.302625],\n },\n \"relevance\": 1,\n \"place_name\": \"Peterborough County, Ontario, Canada\",\n \"place_type\": [\"district\"],\n \"properties\": {\n \"wikidata\": \"Q730542\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFV1lu\",\n },\n },\n {\n \"id\": \"region.17447\",\n \"bbox\": [-95.158717, 41.6400784, -74.3381353, 56.9091182],\n \"text\": \"Ontario\",\n \"type\": \"Feature\",\n \"center\": [-84.2539813003493, 49.4515636581924],\n \"context\": [\n {\n \"id\": \"country.8743\",\n \"text\": \"Canada\",\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"short_code\": \"ca\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [-84.2539813003493, 49.4515636581924],\n },\n \"relevance\": 1,\n \"place_name\": \"Ontario, Canada\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q1904\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpSQ2M\",\n \"short_code\": \"CA-ON\",\n },\n },\n {\n \"id\": \"country.8743\",\n \"bbox\": [-141.010465, 41.6400784, -52.5210105, 83.1472069],\n \"text\": \"Canada\",\n \"type\": \"Feature\",\n \"center\": [-105.750595856519, 55.5859012851966],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [-105.750595856519, 55.5859012851966],\n },\n \"relevance\": 1,\n \"place_name\": \"Canada\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"short_code\": \"ca\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Fire Route 47 K0L 2H0 Lakefield Peterborough County Ontario Canada\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"natural light\",\n \"open area\",\n \"vegetation\",\n \"foliage\",\n \"trees\",\n \"leaves\",\n \"grass\",\n \"man made\",\n \"sunny\",\n ],\n \"categories\": [\"picnic area\"],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"picnic area , outdoor\",\n \"search_location\": \"Fire Route 47 K0L 2H0 Lakefield Peterborough County Ontario Canada\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n 0.07200643420219421,\n -0.27314072847366333,\n -0.28608414530754090,\n -0.017477907240390778,\n 0.50455182790756230,\n 0.28596064448356630,\n 0.06314943730831146,\n 0.23560704290866852,\n -0.19991967082023620,\n 0.05301810801029205,\n 0.23392733931541443,\n -0.015628136694431305,\n -0.12933245301246643,\n -0.62648004293441770,\n 0.19849593937397003,\n -0.09278710186481476,\n -1.13199305534362800,\n 0.02051648497581482,\n 0.18314598500728607,\n -0.07170593738555908,\n -0.35629284381866455,\n 0.10626928508281708,\n 0.022010430693626404,\n -0.36885976791381836,\n 0.01585569977760315,\n 0.19657053053379060,\n -0.33990219235420227,\n -0.15209683775901794,\n -0.13219691812992096,\n -0.29297292232513430,\n -0.007032429799437523,\n 0.30475685000419617,\n -0.69666099548339840,\n -0.55377709865570070,\n 0.50226724147796630,\n 0.43203988671302795,\n -0.03661022335290909,\n 0.12110148370265960,\n 0.27576541900634766,\n -0.51852631568908690,\n -0.16946467757225037,\n 0.06092119216918945,\n -0.05353017896413803,\n -0.10094764083623886,\n 0.24288514256477356,\n -2.32672357559204100,\n 0.23213247954845428,\n -0.31283071637153625,\n 0.05823267251253128,\n 0.033336833119392395,\n 0.03891675919294357,\n 0.34450155496597290,\n 0.54696977138519290,\n -0.10590277612209320,\n -0.38763791322708130,\n 0.11399775743484497,\n -0.11230312287807465,\n -0.25931385159492490,\n -0.71149158477783200,\n 0.19414122402668000,\n -1.06821513175964360,\n -0.05036032199859619,\n 0.23391188681125640,\n 0.18708521127700806,\n -0.28283250331878660,\n -0.49732804298400880,\n 0.14408025145530700,\n 1.58359146118164060,\n -0.09503769129514694,\n 0.10219173878431320,\n 0.18811975419521332,\n 0.11770723760128021,\n 0.14510244131088257,\n -0.025733187794685364,\n -0.12114733457565308,\n 0.028954997658729553,\n -0.36884146928787230,\n -0.14654713869094850,\n -0.20010140538215637,\n 0.05988488718867302,\n 0.02569446712732315,\n 0.52280324697494510,\n -0.00884101539850235,\n -0.84105598926544190,\n 0.08269304037094116,\n -0.030198069289326668,\n 0.08129344880580902,\n -0.49774390459060670,\n -0.61459028720855710,\n -0.27356880903244020,\n -0.24041604995727540,\n -0.33665394783020020,\n -5.96149921417236300,\n 0.22716827690601350,\n 0.28753006458282470,\n 0.48320820927619934,\n -0.12874460220336914,\n 0.14355051517486572,\n -0.34171727299690247,\n -1.38677775859832760,\n 0.058781594038009644,\n 0.09329684078693390,\n 0.12113958597183228,\n 0.02105279266834259,\n 0.25320750474929810,\n 0.30670967698097230,\n -0.23736193776130676,\n -0.008005078881978989,\n -0.15708926320075990,\n -0.11563684791326523,\n -0.19381190836429596,\n -0.70084750652313230,\n -0.25472760200500490,\n -0.25218990445137024,\n 0.04032253846526146,\n 0.15879295766353607,\n -0.24767071008682250,\n 0.28805834054946900,\n 0.02391856536269188,\n 0.018087051808834076,\n 0.21582140028476715,\n 0.62899506092071530,\n -0.13227358460426330,\n 0.30459716916084290,\n 0.08614320307970047,\n -0.23780769109725952,\n 0.25391614437103270,\n -0.22144922614097595,\n -0.08764651417732239,\n 0.37465840578079224,\n -0.19986276328563690,\n -0.23993769288063050,\n -0.24029092490673065,\n 0.86103123426437380,\n -0.20356409251689910,\n 0.18121457099914550,\n 0.007383421063423157,\n 0.02566760778427124,\n -0.18797449767589570,\n 0.12505835294723510,\n 0.06744767725467682,\n -0.31292539834976196,\n 0.14250934123992920,\n -0.16580671072006226,\n -0.10141041129827500,\n 0.07421887665987015,\n 0.13718751072883606,\n 0.31398600339889526,\n -0.45369070768356323,\n 0.37390136718750000,\n -0.17167736589908600,\n -0.12102013081312180,\n 0.55987459421157840,\n -0.64511901140213010,\n 0.32400843501091003,\n -0.15718287229537964,\n 0.03728494048118591,\n -0.48887744545936584,\n -0.43595361709594727,\n 0.27848058938980100,\n -0.26770013570785520,\n 0.11762751638889313,\n 0.14060628414154053,\n 0.14021927118301392,\n 0.16482727229595184,\n 0.12399593740701675,\n 1.10899972915649410,\n 0.58323627710342410,\n 0.25726687908172610,\n -0.04836788773536682,\n 0.42536145448684690,\n 0.026311412453651428,\n 0.23059077560901642,\n -0.09234851598739624,\n -0.42488861083984375,\n -0.52500975131988530,\n -0.14144848287105560,\n 0.031238600611686707,\n -0.22408963739871980,\n 0.16041082143783570,\n 0.19980202615261078,\n 0.12246149778366089,\n -0.74750423431396480,\n 0.34143632650375366,\n -0.07465202361345291,\n 0.032671473920345306,\n 0.35502630472183230,\n -0.54294133186340330,\n 0.01227620244026184,\n 0.11296307295560837,\n 0.23945194482803345,\n 0.22230161726474762,\n 0.09851396828889847,\n -0.10145194828510284,\n -0.94168329238891600,\n -0.63558840751647950,\n -0.09790701419115067,\n 0.08268978446722030,\n 1.47379052639007570,\n -0.40544277429580690,\n -0.19830146431922913,\n 0.12428025901317596,\n 0.01359538733959198,\n 0.11815530061721802,\n -0.14290475845336914,\n -0.25821614265441895,\n -0.27840244770050050,\n 0.51778000593185420,\n -0.28622409701347350,\n 0.13844540715217590,\n 0.019464679062366486,\n -0.64047700166702270,\n -0.51729387044906620,\n 0.36219158768653870,\n -0.29284214973449707,\n -0.09910607337951660,\n -0.44586786627769470,\n 0.26939472556114197,\n 0.001346886157989502,\n 0.10100428760051727,\n 0.12059158086776733,\n 1.46603417396545400,\n -0.16158343851566315,\n 0.09549902379512787,\n 0.31741645932197570,\n -0.06466329097747803,\n -0.37618625164031980,\n -0.10769309103488922,\n 0.24339321255683900,\n -0.07732771337032318,\n -0.28822788596153260,\n -0.30479827523231506,\n 0.28579634428024290,\n 0.57900738716125490,\n 0.01886954903602600,\n -0.34693476557731630,\n 0.13962632417678833,\n -0.36294281482696533,\n -0.10817342996597290,\n 0.035561710596084595,\n -0.57529795169830320,\n -0.12815028429031372,\n 0.03226613253355026,\n -0.58416968584060670,\n 0.13841366767883300,\n 0.03650939464569092,\n 0.038980357348918915,\n 0.34951549768447876,\n 0.20841777324676514,\n 0.12865397334098816,\n -0.12616679072380066,\n 0.08742494136095047,\n 0.21315385401248932,\n 0.09627585113048553,\n 0.20098021626472473,\n -0.06143351271748543,\n -0.043563079088926315,\n -0.19623640179634094,\n 1.57240605354309080,\n -0.35777574777603150,\n 0.02914358116686344,\n 0.63118875026702880,\n 0.18993435800075530,\n -0.46280235052108765,\n 0.030418217182159424,\n -0.34837594628334045,\n 0.03651222586631775,\n 0.26378205418586730,\n 0.046745263040065765,\n 0.28378057479858400,\n -0.009584896266460419,\n 0.19612582027912140,\n 0.21172988414764404,\n -0.032120171934366226,\n 0.20071959495544434,\n -0.08056892454624176,\n 0.12477563321590424,\n -0.13512735068798065,\n 0.14138039946556090,\n 0.31706815958023070,\n 0.024455122649669647,\n -0.15259601175785065,\n -0.20282268524169922,\n -0.28921997547149660,\n 0.45865276455879210,\n 0.11363479495048523,\n -0.20601873099803925,\n 0.12293325364589691,\n -0.007529497146606445,\n -0.20519900321960450,\n -0.27892541885375977,\n -0.25480884313583374,\n 0.10527120530605316,\n -0.22468784451484680,\n 0.27766802906990050,\n 0.27707856893539430,\n 0.42828142642974854,\n -0.06466045975685120,\n -0.022840768098831177,\n 0.08824332058429718,\n 0.13453596830368042,\n 0.21060699224472046,\n -0.13369843363761902,\n -0.42310655117034910,\n 0.06692334264516830,\n 0.018675953149795532,\n -0.012705937027931213,\n 0.16295075416564941,\n -0.006689056754112244,\n -0.27341771125793457,\n -0.45861658453941345,\n 0.14121453464031220,\n 0.85903668403625490,\n -0.19725802540779114,\n 0.14617547392845154,\n 0.27198433876037600,\n 0.17599108815193176,\n 0.02266123890876770,\n -0.12822464108467102,\n 1.00072443485260000,\n 0.30147033929824830,\n -0.74144268035888670,\n -0.16088382899761200,\n 0.11513932049274445,\n -0.38404512405395510,\n 0.04975423216819763,\n 0.37200531363487244,\n 0.19585512578487396,\n -0.12441303580999374,\n -0.24220986664295197,\n 0.15007376670837402,\n -0.04617227613925934,\n -0.04232384264469147,\n -0.34620440006256104,\n -0.26701658964157104,\n -0.005242077633738518,\n 0.0031939297914505005,\n 0.02201770246028900,\n -0.09960493445396423,\n 0.16118554770946503,\n -0.33185213804244995,\n 0.15700313448905945,\n 0.16201999783515930,\n 0.07118913531303406,\n 0.22056972980499268,\n -0.011956900358200073,\n -0.0003955662250518799,\n 0.25377553701400757,\n -0.41737711429595950,\n 0.26137113571166990,\n 0.42841473221778870,\n -0.11275280267000198,\n -0.54838216304779050,\n 0.03743823617696762,\n 0.08181229233741760,\n 1.28952217102050780,\n 0.36465561389923096,\n -0.28938600420951843,\n 0.47177675366401670,\n -0.16908544301986694,\n 0.93969750404357910,\n 0.08721843361854553,\n -0.09430234134197235,\n 0.66503000259399410,\n -0.48145574331283570,\n -0.14018303155899048,\n 0.33783870935440063,\n -0.75327533483505250,\n -0.52762186527252200,\n 0.04408954456448555,\n -0.39755284786224365,\n -0.0019524451345205307,\n 0.30438941717147827,\n 0.04218447208404541,\n -0.07897476106882095,\n 0.35774332284927370,\n 1.22708070278167720,\n 0.29904451966285706,\n -0.11844162642955780,\n -0.15679849684238434,\n -0.12561301887035370,\n -0.77952718734741210,\n -0.26338329911231995,\n -0.08969837427139282,\n 0.21005888283252716,\n -0.31030485033988950,\n 0.24320170283317566,\n 0.27376961708068850,\n -0.18231861293315887,\n -1.11225414276123050,\n -0.96022212505340580,\n 0.39804935455322266,\n 0.007329510059207678,\n -0.21168775856494904,\n -0.63149207830429080,\n 0.21230995655059814,\n 0.13413298130035400,\n 0.38083824515342710,\n 0.21768637001514435,\n -0.37690994143486023,\n 0.010201409459114075,\n 0.35002541542053220,\n -0.00883527286350727,\n -0.49314117431640625,\n 0.11570226401090622,\n -0.12012594938278198,\n 0.14163519442081451,\n -0.31398567557334900,\n -0.11759512126445770,\n -0.26619073748588560,\n 0.24936842918395996,\n -0.07021093368530273,\n -0.26329314708709717,\n -0.26164028048515320,\n -0.48863250017166140,\n -0.21155467629432678,\n -0.51766830682754520,\n -0.41783249378204346,\n -0.36551034450531006,\n -0.69736742973327640,\n 0.33944544196128845,\n -0.14483065903186798,\n 0.44479277729988100,\n -0.98320567607879640,\n -0.19574180245399475,\n 0.31662541627883910,\n -0.44551515579223633,\n -0.022557497024536133,\n -0.48501241207122800,\n -0.09746624529361725,\n -0.36100867390632630,\n -0.27448734641075134,\n 0.78780412673950200,\n -0.30192264914512634,\n -0.44880360364913940,\n -0.033154018223285675,\n 0.048185281455516815,\n 0.03699145466089249,\n 0.46661505103111267,\n -0.06928810477256775,\n 0.32975050806999207,\n 0.66338658332824710,\n 0.026022130623459816,\n -0.02559243142604828,\n 0.13531213998794556,\n -0.011617228388786316,\n -0.24844856560230255,\n -0.22130966186523438,\n 0.06896549463272095,\n 0.16910549998283386,\n -0.33283072710037230,\n -0.23490653932094574,\n -0.42383372783660890,\n -0.82607471942901610,\n -0.12491527199745178,\n 0.09555611014366150,\n 0.14898100495338440,\n 0.41498780250549316,\n -0.33365392684936523,\n 0.08857516199350357,\n -0.25846809148788450,\n -0.023531511425971985,\n 0.24037255346775055,\n 0.25419849157333374,\n 0.18990617990493774,\n 0.21707595884799957,\n 0.054550834000110626,\n -0.16628548502922058,\n 0.29191961884498596,\n 0.21750520169734955,\n 0.059166401624679565,\n 0.27432167530059814,\n -0.07397265732288360,\n -0.39545136690139770,\n 0.32126304507255554,\n -0.23423250019550323,\n -0.46515718102455140,\n 0.23323012888431550,\n 0.22779177129268646,\n -0.19990277290344238,\n 0.16021879017353058,\n 0.12057007849216461,\n 0.48086464405059814,\n 0.21143439412117004,\n -0.13717344403266907,\n -0.26784723997116090,\n 0.36808949708938600,\n 0.04166709631681442,\n 0.10163601487874985,\n -0.65290242433547970,\n 0.001865200698375702,\n 0.24058419466018677,\n 0.43577551841735840,\n -0.30626660585403440,\n 0.04026487469673157,\n -0.59185659885406490,\n 0.28134933114051820,\n 0.10597439110279083,\n -0.02836729586124420,\n -0.17641222476959229,\n -0.41398856043815613,\n 0.13548584282398224,\n 0.03822256624698639,\n 0.49377080798149110,\n 0.79975557327270510,\n -0.52094459533691410,\n 0.37935078144073486,\n -0.22603368759155273,\n -0.30841892957687380,\n 0.0037401914596557617,\n -0.19617760181427002,\n -0.03347594290971756,\n ],\n \"clip_embeddings_magnitude\": 10.361903190612793,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[72, 81, 54]\",\n \"video_length\": None,\n \"removed\": False,\n \"in_trashcan\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"1/120\",\n \"size\": 12186941,\n \"subjectDistance\": 3.91,\n \"width\": 4032,\n \"main_file_id\": \"c4d896c9c9c0ca3c312b19f2f22d69ae1\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/e21b9ade718afdbe931462d837b74d1c1.webp\",\n \"square_thumbnail\": \"square_thumbnails/e21b9ade718afdbe931462d837b74d1c1.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/e21b9ade718afdbe931462d837b74d1c1.webp\",\n \"added_on\": \"2023-06-16 16:30:55.449974 +00:00\",\n \"exif_gps_lat\": 52.5019361111111,\n \"exif_gps_lon\": 13.3815138888889,\n \"exif_timestamp\": \"2017-08-22 13:27:09.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [13.381514, 52.501936],\n \"features\": [\n {\n \"id\": \"poi.1047972024770\",\n \"text\": \"Roncalli\",\n \"type\": \"Feature\",\n \"center\": [13.381502, 52.501711],\n \"context\": [\n {\n \"id\": \"postcode.5541434\",\n \"text\": \"10963\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpWSTQ2\",\n },\n {\n \"id\": \"locality.181160506\",\n \"text\": \"Kreuzberg\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpDc3hLT2c\",\n },\n {\n \"id\": \"place.115770\",\n \"text\": \"Berlin\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.381502, 52.501711],\n },\n \"relevance\": 1,\n \"place_name\": \"Roncalli, Möckernstr, Berlin, 10963, Germany\",\n \"place_type\": [\"poi\"],\n \"properties\": {\n \"maki\": \"theatre\",\n \"address\": \"Möckernstr\",\n \"category\": \"circus\",\n \"landmark\": True,\n \"foursquare\": \"50d22a90e4b0d3035504ae2b\",\n },\n },\n {\n \"id\": \"postcode.5541434\",\n \"bbox\": [13.372592, 52.49205, 13.391193, 52.508285],\n \"text\": \"10963\",\n \"type\": \"Feature\",\n \"center\": [13.383894, 52.497062],\n \"context\": [\n {\n \"id\": \"locality.181160506\",\n \"text\": \"Kreuzberg\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpDc3hLT2c\",\n },\n {\n \"id\": \"place.115770\",\n \"text\": \"Berlin\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.383894, 52.497062],\n },\n \"relevance\": 1,\n \"place_name\": \"10963, Berlin, Germany\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpWSTQ2\"},\n },\n {\n \"id\": \"locality.181160506\",\n \"bbox\": [13.368215, 52.48277, 13.453351, 52.50938],\n \"text\": \"Kreuzberg\",\n \"type\": \"Feature\",\n \"center\": [13.411914, 52.497644],\n \"context\": [\n {\n \"id\": \"place.115770\",\n \"text\": \"Berlin\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.411914, 52.497644],\n },\n \"relevance\": 1,\n \"place_name\": \"Kreuzberg, Berlin, Germany\",\n \"place_type\": [\"locality\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpDc3hLT2c\"},\n },\n {\n \"id\": \"place.115770\",\n \"bbox\": [13.08836, 52.338261, 13.761131, 52.675502],\n \"text\": \"Berlin\",\n \"type\": \"Feature\",\n \"center\": [13.3888599, 52.5170365],\n \"context\": [\n {\n \"id\": \"country.8762\",\n \"text\": \"Germany\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [13.3888599, 52.5170365],\n },\n \"relevance\": 1,\n \"place_name\": \"Berlin, Germany\",\n \"place_type\": [\"region\", \"place\"],\n \"properties\": {\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"short_code\": \"DE-BE\",\n },\n },\n {\n \"id\": \"country.8762\",\n \"bbox\": [5.866315, 47.270238, 15.041832, 55.1286491],\n \"text\": \"Germany\",\n \"type\": \"Feature\",\n \"center\": [10.0183432948567, 51.1334813439932],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [10.0183432948567, 51.1334813439932],\n },\n \"relevance\": 1,\n \"place_name\": \"Germany\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"short_code\": \"de\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Roncalli 10963 Kreuzberg Berlin Germany\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"natural light\",\n \"open area\",\n \"man made\",\n \"clouds\",\n \"no horizon\",\n \"sunny\",\n \"far away horizon\",\n \"cloth\",\n \"cold\",\n ],\n \"categories\": [],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"outdoor\",\n \"search_location\": \"Roncalli 10963 Kreuzberg Berlin Germany\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n -0.10258262604475021,\n 0.66643309593200680,\n -0.06443142890930176,\n -0.14745938777923584,\n 0.48618343472480774,\n 0.0004399120807647705,\n -0.0009457916021347046,\n 0.64925503730773930,\n -0.45721071958541870,\n 0.11818008124828339,\n -0.11370902508497238,\n -0.49487552046775820,\n -0.28128325939178467,\n -0.30589711666107180,\n -0.15134385228157043,\n 0.05521773546934128,\n -1.55117774009704600,\n -0.16058634221553802,\n -0.13487541675567627,\n -0.15009154379367828,\n -0.34108334779739380,\n -0.13820111751556396,\n 0.09468824416399002,\n 0.34886407852172850,\n -0.34096065163612366,\n -0.02098141610622406,\n -0.17523168027400970,\n -0.052726227790117264,\n 0.25815361738204956,\n -0.45174264907836914,\n -0.63614970445632930,\n 0.09681782126426697,\n -0.11660185456275940,\n -0.18351450562477112,\n 0.41413304209709170,\n 0.14657795429229736,\n -0.28526934981346130,\n -0.22788083553314210,\n -0.21612341701984406,\n -0.73998987674713130,\n -0.55964195728302000,\n -0.22357036173343658,\n -0.22361208498477936,\n 0.002233438193798065,\n 0.50072944164276120,\n -1.46298265457153320,\n 0.63237798213958740,\n 0.03941209241747856,\n 0.19379198551177979,\n -0.44258201122283936,\n -0.17907603085041046,\n 0.06887031346559525,\n 0.51105743646621700,\n 0.25896197557449340,\n 0.010099773295223713,\n 0.60815042257308960,\n -0.21678172051906586,\n 0.22954954206943512,\n -0.71181941032409670,\n 0.17145885527133942,\n 0.79170489311218260,\n -0.07573703676462173,\n 0.25271376967430115,\n 0.07294708490371704,\n 0.08441808819770813,\n 0.27773401141166687,\n -0.10114270448684692,\n 1.18037462234497070,\n -0.10031361877918243,\n -0.57329255342483520,\n 0.15199229121208190,\n 0.08043078333139420,\n -0.25711309909820557,\n 0.35362368822097780,\n 0.19381424784660340,\n 0.07242968678474426,\n -0.30958822369575500,\n -0.08953941613435745,\n -0.09282187372446060,\n -0.37684765458106995,\n -0.31265074014663696,\n 0.76198458671569820,\n -0.24638760089874268,\n -0.18398816883563995,\n 0.20948600769042970,\n 0.68236070871353150,\n 0.26739034056663513,\n -0.19240638613700867,\n -0.31598865985870360,\n -0.029354378581047058,\n 0.008327040821313858,\n 0.18494442105293274,\n -7.53693056106567400,\n 0.25613960623741150,\n 0.22370323538780212,\n -0.03313902020454407,\n -0.22441212832927704,\n 0.24500671029090880,\n -0.67216014862060550,\n -0.63962388038635250,\n -0.10790157318115234,\n -0.39841800928115845,\n 0.45670357346534730,\n 0.07113364338874817,\n -0.28614389896392820,\n 0.35553669929504395,\n -1.00858056545257570,\n 0.17648085951805115,\n -0.56130737066268920,\n 0.33491787314414980,\n 0.27355626225471497,\n 0.16742312908172607,\n 0.36952716112136840,\n -0.32609111070632935,\n -0.21147111058235168,\n -0.09825261682271957,\n -0.47114884853363037,\n 0.60953688621521000,\n 0.38453266024589540,\n -0.38745057582855225,\n -0.04564698040485382,\n 0.06692779809236526,\n 0.007835905998945236,\n 0.12590317428112030,\n 0.08536458760499954,\n -0.22401364147663116,\n 0.15795263648033142,\n -0.46122899651527405,\n -0.06393150240182877,\n 0.27560144662857056,\n 0.29461151361465454,\n -0.22232146561145782,\n -0.14870893955230713,\n 1.00925290584564200,\n 0.08114692568778992,\n 0.010658219456672668,\n 0.019297055900096893,\n -0.37022376060485840,\n -0.60042166709899900,\n -0.08197715133428574,\n 0.17349836230278015,\n -0.40405350923538210,\n 0.05475588142871857,\n 0.64725798368453980,\n -0.19707715511322021,\n 0.10167735815048218,\n -0.24003769457340240,\n 0.47526404261589050,\n -0.17396590113639832,\n 0.52764558792114260,\n -0.30993252992630005,\n -0.06276136636734009,\n 0.54091024398803710,\n -0.32085520029067993,\n 0.06873718649148941,\n -0.79230988025665280,\n 0.12832219898700714,\n 0.27965274453163147,\n -0.11812235414981842,\n -0.025007829070091248,\n -0.05913873016834259,\n 0.009386129677295685,\n 0.74419283866882320,\n 0.53745734691619870,\n -0.07048669457435608,\n 0.32115909457206726,\n 0.85237395763397220,\n -0.007245570421218872,\n -0.20299741625785828,\n -0.31570893526077270,\n 0.36141934990882874,\n 0.026256147772073746,\n -0.20647422969341278,\n 0.07242944836616516,\n -0.39613524079322815,\n -0.15577512979507446,\n 0.56382685899734500,\n 0.15884301066398620,\n -0.25236248970031740,\n 0.12900674343109130,\n 0.13617977499961853,\n -0.38444459438323975,\n -0.17406651377677917,\n 0.38698321580886840,\n -0.03219318389892578,\n 0.22693467140197754,\n 0.061593152582645416,\n -0.21263362467288970,\n 0.13701051473617554,\n -0.057763803750276566,\n 0.25249591469764710,\n 0.08275345712900162,\n 0.20910006761550903,\n 0.03912349045276642,\n -0.63464707136154170,\n -0.30452278256416320,\n -0.32845041155815125,\n -0.06514476239681244,\n 0.37253341078758240,\n 0.27116832137107850,\n 0.08592364937067032,\n 0.23760634660720825,\n -0.28635382652282715,\n -0.04585602134466171,\n -0.25701779127120970,\n -0.18018849194049835,\n -0.48961794376373290,\n -0.08290880918502808,\n -0.33540713787078860,\n 0.07882565259933472,\n 0.39085230231285095,\n 0.05874055624008179,\n 0.057173386216163635,\n -0.08757130801677704,\n 0.07342680543661118,\n -0.13899344205856323,\n -0.08424560725688934,\n 0.19400805234909058,\n -0.72243881225585940,\n -0.29955309629440310,\n 0.20382502675056458,\n 0.48866674304008484,\n 0.07971449941396713,\n -0.03938521072268486,\n -0.62366402149200440,\n -0.15463593602180480,\n 0.09072853624820709,\n 0.03131377696990967,\n 0.28323334455490110,\n -0.16908100247383118,\n -0.20017117261886597,\n 0.40308171510696410,\n 0.45233413577079773,\n -0.31968781352043150,\n -0.14290061593055725,\n -0.73431730270385740,\n 0.32717624306678770,\n -0.24733215570449830,\n 0.07128322124481201,\n -0.48950380086898804,\n -0.0008872002363204956,\n -0.07294179499149323,\n -0.22609096765518188,\n -0.020967215299606323,\n -0.07549750804901123,\n 0.20787967741489410,\n 0.36314973235130310,\n 0.12670208513736725,\n -0.31853759288787840,\n 0.32616245746612550,\n -0.09472109377384186,\n -0.027377739548683167,\n 0.014935225248336792,\n 0.20435945689678192,\n -0.16135276854038239,\n 0.19143480062484740,\n 0.08755904436111450,\n 0.20116862654685974,\n -0.08494839072227478,\n -0.22879388928413390,\n 0.22606262564659120,\n -0.03982345014810562,\n -0.29927518963813780,\n 0.010079406201839447,\n 0.07509651780128479,\n -0.25567504763603210,\n -0.11709715425968170,\n 0.42464250326156616,\n 0.06431484222412110,\n 0.17397393286228180,\n 0.28166836500167847,\n -0.21547360718250275,\n -0.06485859304666519,\n -0.07584954798221588,\n 0.12973758578300476,\n -0.06666970998048782,\n -0.19354835152626038,\n -0.27083098888397217,\n -0.08537831902503967,\n -0.24621208012104034,\n -0.11692193150520325,\n 0.12622721493244170,\n 0.64608305692672730,\n 0.07786636054515839,\n -0.15758465230464935,\n 0.15170501172542572,\n 0.14211302995681763,\n -0.31157767772674560,\n -0.15162092447280884,\n -0.10263395309448242,\n 0.13374274969100952,\n -0.35772666335105896,\n 0.16042037308216095,\n -0.09574175626039505,\n 0.13830652832984924,\n 0.33334568142890930,\n -0.53578650951385500,\n 0.39855188131332400,\n -0.37783429026603700,\n -0.25219041109085083,\n 0.06827493011951447,\n -0.50635170936584470,\n -0.13055899739265442,\n 0.050748057663440704,\n 0.14656113088130950,\n -0.85355234146118160,\n -0.03550938516855240,\n 0.21138493716716766,\n 0.51412421464920040,\n -0.16174469888210297,\n 0.06829738616943360,\n 0.38762015104293823,\n 1.00867974758148200,\n -0.11224977672100067,\n 0.23601248860359192,\n 0.21366406977176666,\n 0.22318972647190094,\n -0.57790690660476680,\n -0.20985898375511170,\n -0.13206046819686890,\n 0.43933963775634766,\n -0.76506268978118900,\n -0.33175840973854065,\n 0.35506504774093630,\n -0.66501140594482420,\n 0.40778365731239320,\n -0.19959415495395660,\n -0.28629922866821290,\n 0.27318397164344790,\n 0.37771254777908325,\n -0.12135715782642365,\n 0.011033750139176846,\n -0.05133216455578804,\n -0.06439136713743210,\n -0.07881140708923340,\n -0.07496929168701172,\n 0.36653387546539307,\n -0.11792770028114319,\n 0.33755394816398620,\n -0.09257731586694717,\n 0.009545769542455673,\n -0.04660305380821228,\n 0.013396577909588814,\n 0.15035420656204224,\n -0.61903655529022220,\n -0.17708726227283478,\n -0.013831155374646187,\n 0.010356694459915161,\n 0.10595686733722687,\n -0.04540724307298660,\n 0.0009019002318382263,\n -0.26157030463218690,\n 0.18670830130577087,\n -0.059031642973423004,\n -0.11484433710575104,\n -0.36731621623039246,\n 0.14339649677276610,\n 1.26058232784271240,\n -0.015540368854999542,\n -0.07888511568307877,\n -0.009377516806125641,\n -0.15602213144302368,\n -0.12761303782463074,\n 0.25213027000427246,\n 0.92153555154800420,\n -0.18542146682739258,\n -0.21386396884918213,\n -1.57489061355590820,\n 0.014670997858047485,\n 0.35590082406997680,\n 0.23226591944694520,\n 0.29403144121170044,\n 0.19039350748062134,\n 0.12028850615024567,\n 0.14649519324302673,\n -0.57438409328460690,\n 1.51094436645507810,\n -0.06900059431791306,\n -0.09223778545856476,\n 0.05437563359737396,\n 0.12748995423316956,\n -0.07933007180690765,\n -0.81089299917221070,\n -0.47452667355537415,\n 0.12268605083227158,\n 0.40242570638656616,\n 0.28311154246330260,\n 0.27250385284423830,\n -0.23757700622081757,\n 0.32199540734291077,\n -0.31787779927253723,\n 0.24869447946548462,\n -0.32091945409774780,\n 0.13392969965934753,\n 0.024256711825728416,\n -0.003996938467025757,\n 0.31115883588790894,\n -0.50246292352676390,\n -0.20639395713806152,\n 0.03759238123893738,\n 0.17185325920581818,\n -0.030085332691669464,\n 0.048362910747528076,\n -0.62065368890762330,\n -0.020879443734884262,\n 0.12434196472167969,\n 0.040695954114198685,\n -0.50741922855377200,\n 0.08081452548503876,\n 0.14095300436019897,\n -0.15079917013645172,\n 0.14225539565086365,\n -0.03339875489473343,\n 0.25852018594741820,\n -0.41237697005271910,\n 0.31924033164978030,\n -0.25935268402099610,\n -0.32131356000900270,\n -0.69814765453338620,\n -0.22040528059005737,\n 0.05100163817405701,\n -0.15426829457283020,\n 0.22865183651447296,\n -0.47168889641761780,\n -0.31051588058471680,\n 0.048167359083890915,\n -0.21201044321060180,\n 0.10385461151599884,\n 0.18401621282100677,\n 0.12361292541027069,\n 0.30423548817634580,\n -0.17449685931205750,\n 0.31314414739608765,\n 0.04636995494365692,\n -0.41227555274963380,\n 0.045520931482315063,\n 0.53545415401458740,\n 0.13680092990398407,\n -0.36478853225708010,\n 0.40942415595054626,\n 0.23506653308868408,\n 1.04005861282348630,\n -0.27912554144859314,\n 0.06439234316349030,\n 0.18784391880035400,\n -0.17379683256149292,\n -0.18675222992897034,\n -0.34193059802055360,\n -0.38465169072151184,\n 0.34344249963760376,\n 0.23465913534164430,\n 0.032892048358917236,\n -0.28188407421112060,\n -0.11934018135070801,\n 0.022270366549491882,\n 0.07610686123371124,\n 0.16484493017196655,\n 0.28013628721237180,\n -0.06442663818597794,\n -0.13154838979244232,\n 0.38454037904739380,\n -0.15238852798938750,\n -0.34664541482925415,\n 0.06196160987019539,\n 0.21366979181766510,\n -0.23718816041946410,\n 0.29661375284194946,\n 0.07324250787496567,\n 0.02359824627637863,\n -0.29029119014739990,\n -0.04170669615268707,\n -0.13716265559196472,\n -0.18890701234340668,\n -0.71853756904602050,\n -0.14910481870174408,\n -0.28566592931747437,\n 0.17772153019905090,\n 0.08817628026008606,\n -0.034255996346473694,\n -0.23667897284030914,\n -0.62015527486801150,\n 0.20481406152248383,\n 0.19941198825836182,\n -0.11482881009578705,\n 0.006309971213340759,\n 0.42098861932754517,\n 0.03637159615755081,\n 0.42487934231758120,\n -0.13934084773063660,\n -0.32979014515876770,\n 0.13939698040485382,\n 0.15597808361053467,\n 0.18135486543178558,\n -0.73175537586212160,\n -0.26711130142211914,\n -0.09930166602134705,\n -0.30019181966781616,\n 0.27222818136215210,\n -0.13047066330909730,\n 0.21958050131797790,\n -0.92940568923950200,\n 0.46761587262153625,\n 0.002422422170639038,\n 0.06002846360206604,\n -0.07387303560972214,\n -0.14259713888168335,\n -0.52521455287933350,\n 0.21675816178321838,\n -0.26278591156005860,\n -0.029557587578892708,\n -0.39842849969863890,\n -0.21753847599029540,\n ],\n \"clip_embeddings_magnitude\": 10.904964447021484,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[255, 255, 255]\",\n \"video_length\": None,\n \"in_trashcan\": False,\n \"removed\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.441,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"0\",\n \"size\": 4303517,\n \"subjectDistance\": 1.356,\n \"width\": 4032,\n \"main_file_id\": \"e21b9ade718afdbe931462d837b74d1c1\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/a5bcc802a708c07c18816d1b480a8b7d1.webp\",\n \"square_thumbnail\": \"square_thumbnails/a5bcc802a708c07c18816d1b480a8b7d1.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/a5bcc802a708c07c18816d1b480a8b7d1.webp\",\n \"added_on\": \"2023-06-16 16:30:52.549542 +00:00\",\n \"exif_gps_lat\": -33.9432611111111,\n \"exif_gps_lon\": 151.263833333333,\n \"exif_timestamp\": \"2017-08-17 09:31:53.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [151.263833, -33.943261],\n \"features\": [\n {\n \"id\": \"poi.695784709457\",\n \"text\": \"Mahon Rock Pool\",\n \"type\": \"Feature\",\n \"center\": [151.263303, -33.94292],\n \"context\": [\n {\n \"id\": \"locality.269412878\",\n \"text\": \"Maroubra\",\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.263303, -33.94292],\n },\n \"relevance\": 1,\n \"place_name\": \"Mahon Rock Pool, Marine Pde., Sydney, New South Wales, Australia\",\n \"place_type\": [\"poi\"],\n \"properties\": {\n \"address\": \"Marine Pde.\",\n \"category\": \"swimming pool, pool, swim club\",\n \"landmark\": True,\n \"foursquare\": \"4b550d36f964a52055d927e3\",\n },\n },\n {\n \"id\": \"locality.269412878\",\n \"bbox\": [151.226804557, -33.958002321, 151.292531935, -33.93283624],\n \"text\": \"Maroubra\",\n \"type\": \"Feature\",\n \"center\": [151.2575, -33.9475],\n \"context\": [\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [151.2575, -33.9475]},\n \"relevance\": 1,\n \"place_name\": \"Maroubra, New South Wales, Australia\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n },\n {\n \"id\": \"place.24496142\",\n \"bbox\": [150.520934139, -34.11717528, 151.369884128, -33.562644328],\n \"text\": \"Sydney\",\n \"type\": \"Feature\",\n \"center\": [151.216454, -33.854816],\n \"context\": [\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.216454, -33.854816],\n },\n \"relevance\": 1,\n \"place_name\": \"Sydney, New South Wales, Australia\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n },\n {\n \"id\": \"region.33806\",\n \"bbox\": [140.999265, -37.5097258, 159.200456, -28.1370359],\n \"text\": \"New South Wales\",\n \"type\": \"Feature\",\n \"center\": [147.014694071448, -32.168971672412],\n \"context\": [\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [147.014694071448, -32.168971672412],\n },\n \"relevance\": 1,\n \"place_name\": \"New South Wales, Australia\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n },\n {\n \"id\": \"country.8718\",\n \"bbox\": [112.8256904, -54.8327658, 159.200456, -9.0436707],\n \"text\": \"Australia\",\n \"type\": \"Feature\",\n \"center\": [134.489562606981, -25.7349684916223],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [134.489562606981, -25.7349684916223],\n },\n \"relevance\": 1,\n \"place_name\": \"Australia\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Mahon Rock Pool Maroubra Sydney New South Wales Australia\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"natural light\",\n \"open area\",\n \"far away horizon\",\n \"natural\",\n \"sunny\",\n \"boating\",\n \"moist\",\n \"swimming\",\n \"ocean\",\n ],\n \"categories\": [\"marsh\", \"tundra\", \"lagoon\"],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"marsh , tundra , lagoon , outdoor\",\n \"search_location\": \"Mahon Rock Pool Maroubra Sydney New South Wales Australia\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n 0.12998622655868530,\n 0.13258025050163270,\n -0.058508455753326416,\n 0.22892819344997406,\n -0.023302078247070312,\n -0.44510543346405030,\n 0.43837052583694460,\n 0.40075856447219850,\n 0.42637223005294800,\n -0.18267586827278137,\n -0.32148724794387820,\n -0.03853927552700043,\n -0.059241898357868195,\n -0.12733995914459229,\n -0.09387104958295822,\n -0.43112272024154663,\n -1.19327056407928470,\n -0.014609143137931824,\n 0.21503102779388428,\n -0.13053396344184875,\n 0.28765681385993960,\n 0.52494764328002930,\n 0.42401337623596190,\n 0.011837676167488098,\n 0.09145589917898178,\n 0.22389806807041168,\n -0.036226868629455566,\n 0.09836312383413315,\n -0.10595801472663880,\n 0.20653992891311646,\n 0.26259601116180420,\n -0.021937429904937744,\n -0.37160769104957580,\n 0.06919395923614502,\n 0.17206385731697083,\n 0.05610951408743858,\n -0.25814598798751830,\n 0.25112861394882200,\n 0.13220319151878357,\n 0.22801768779754640,\n -0.28782862424850464,\n 0.52349138259887700,\n 0.11140240728855133,\n 0.21639062464237213,\n 0.16601549088954926,\n -1.26960456371307370,\n 0.03704804554581642,\n 0.15477925539016724,\n 0.32343593239784240,\n 0.05724513530731201,\n 0.23438473045825958,\n 0.34855538606643677,\n 0.30527845025062560,\n 0.047757647931575775,\n -0.14104431867599487,\n 0.28336334228515625,\n 0.12284316122531891,\n 0.09858018159866333,\n -0.64525383710861210,\n -0.28189745545387270,\n -0.53113412857055660,\n -0.02064698189496994,\n -0.029455430805683136,\n 0.31310641765594480,\n -0.01327805221080780,\n -0.05426792800426483,\n -0.12057886272668839,\n 1.52634048461914060,\n -0.23908194899559020,\n 0.03377249091863632,\n -0.31758821010589600,\n 0.003850087523460388,\n -0.44408404827117920,\n -0.58076667785644530,\n 0.022277936339378357,\n 0.21131259202957153,\n -0.17391380667686462,\n -0.28281235694885254,\n 0.19357018172740936,\n -0.047077856957912445,\n -0.022741809487342834,\n 0.22117993235588074,\n -0.06992056965827942,\n 0.18571671843528748,\n 0.06705777347087860,\n 0.24204009771347046,\n 0.11218206584453583,\n -0.03229340910911560,\n 0.20023432374000550,\n -0.08615788817405700,\n 0.29021203517913820,\n 0.11631032079458237,\n -6.34464693069458000,\n -0.37301033735275270,\n -0.08067938685417175,\n 0.23319250345230103,\n -0.11491980403661728,\n -0.22198167443275452,\n -0.23849581182003020,\n -1.53279995918273930,\n 0.25409278273582460,\n -0.37143617868423460,\n 0.12711408734321594,\n 0.18552611768245697,\n 0.13125161826610565,\n -0.03438793122768402,\n 0.12672650814056396,\n -0.30418828129768370,\n 0.05400300398468971,\n 0.22966152429580688,\n 0.10130165517330170,\n -0.18591997027397156,\n 0.16613908112049103,\n 0.08785743266344070,\n -0.09753852337598800,\n -0.19177854061126710,\n -0.54351925849914550,\n -0.029669322073459625,\n 0.008011367172002792,\n 0.06446681916713715,\n 0.23291164636611938,\n -0.78047883510589600,\n -0.35974910855293274,\n -0.57393378019332890,\n 0.005923539400100708,\n -0.07117588073015213,\n 0.13031338155269623,\n -0.43962341547012330,\n -0.75585317611694340,\n 0.27976679801940920,\n -0.005890890955924988,\n 0.16367265582084656,\n 0.010072827339172363,\n 0.86911255121231080,\n -0.50223058462142940,\n 0.09843732416629791,\n -0.03996431455016136,\n -0.25243335962295530,\n 0.011986486613750458,\n 0.07025257498025894,\n -0.15618172287940980,\n -0.13723298907279968,\n 0.14785155653953552,\n -0.07476790249347687,\n 0.011668689548969269,\n 0.15818738937377930,\n -0.22708828747272491,\n 1.15767204761505130,\n -0.24034819006919860,\n 0.06645347177982330,\n 0.03723381087183952,\n 0.13653486967086792,\n -0.91293966770172120,\n 0.061923325061798096,\n 0.020089279860258102,\n -0.19603200256824493,\n 0.54111683368682860,\n 0.08232687413692474,\n -0.20884832739830017,\n 0.34021350741386414,\n -0.87326097488403320,\n -0.27245533466339110,\n 0.30318510532379150,\n 0.30751353502273560,\n 0.25488168001174927,\n -0.19319717586040497,\n 1.10999667644500730,\n 0.32085287570953370,\n 0.19622525572776794,\n -0.29783028364181520,\n -0.21281230449676514,\n 0.002480059862136841,\n 0.16586032509803772,\n 0.16510939598083496,\n -0.25083601474761963,\n 0.01780378818511963,\n -0.23946386575698853,\n 0.26920223236083984,\n -0.60473775863647460,\n 0.71488064527511600,\n 0.90620148181915280,\n -0.11498169600963593,\n -0.11332808434963226,\n 0.29238957166671753,\n -0.28703141212463380,\n -0.08944821357727051,\n 0.33648857474327090,\n -0.27088621258735657,\n 0.056357257068157196,\n 0.06953722238540650,\n -0.12937217950820923,\n -0.34909275174140930,\n 0.0033053942024707794,\n 0.36867564916610720,\n 0.02737647294998169,\n -0.40904921293258667,\n 0.15042823553085327,\n -0.23831282556056976,\n 1.08098745346069340,\n -0.14632245898246765,\n -0.10184669494628906,\n -0.45605412125587463,\n -0.26751014590263367,\n -0.19427978992462158,\n -0.07291503250598907,\n -0.23394347727298737,\n -0.39399421215057373,\n 0.21800534427165985,\n 0.08165672421455383,\n 0.31118094921112060,\n -0.03596162423491478,\n 0.56992232799530030,\n -0.25306987762451170,\n -0.013975441455841064,\n -0.13325847685337067,\n 0.19483233988285065,\n 0.10027952492237091,\n 0.57050353288650510,\n 0.07020722329616547,\n -0.30141943693161010,\n 0.10414308309555054,\n 0.39053773880004883,\n 0.13395856320858002,\n 0.06220602989196777,\n -0.39614889025688170,\n 0.021074017509818077,\n -0.27187311649322510,\n 0.50517594814300540,\n 0.14774698019027710,\n 0.35505199432373047,\n -0.014332786202430725,\n 0.16666226089000702,\n 0.16399480402469635,\n 0.09917092323303223,\n 0.35650873184204100,\n -1.40380215644836430,\n -0.026044342666864395,\n -0.19442640244960785,\n -0.24114309251308440,\n 0.058344028890132904,\n 0.12207233905792236,\n -0.16192662715911865,\n 0.09302338957786560,\n 0.49859148263931274,\n -0.14120854437351227,\n -0.05409412086009979,\n -0.21078500151634216,\n 0.32789957523345950,\n 0.17087182402610780,\n 0.03868192434310913,\n -0.58889353275299070,\n 0.18010029196739197,\n 0.04883924126625061,\n 0.26769936084747314,\n -0.23523572087287903,\n 0.26934176683425903,\n -0.23642563819885254,\n 0.12310450524091720,\n 0.89556938409805300,\n -0.06872670352458954,\n -0.04263883829116821,\n 0.17438516020774840,\n 0.25022634863853455,\n -0.06414982676506042,\n 0.17062856256961823,\n -0.67373126745224000,\n 0.33244323730468750,\n -0.11403056979179382,\n -0.28086948394775390,\n -0.023052534088492393,\n 0.10679151117801666,\n 0.13404421508312225,\n 0.12281967699527740,\n -0.48176917433738710,\n -0.50348985195159910,\n 0.19038720428943634,\n 0.06314077228307724,\n 0.12262192368507385,\n -0.35703650116920470,\n 0.24681422114372253,\n -0.07084030658006668,\n 0.13866774737834930,\n 0.020786508917808533,\n 0.48234757781028750,\n 0.18154349923133850,\n 0.24811989068984985,\n -0.47986146807670593,\n -0.26600340008735657,\n -0.79610115289688110,\n -0.44987118244171140,\n 0.03849541395902634,\n -0.26184004545211790,\n 0.42152988910675050,\n -0.30942755937576294,\n 0.14844314754009247,\n -0.12019725143909454,\n 1.18861508369445800,\n 0.02964150905609131,\n -0.38321954011917114,\n -0.17666542530059814,\n -0.24750977754592896,\n -0.18742603063583374,\n 0.23742599785327910,\n -0.13876700401306152,\n 0.53921639919281010,\n 0.28352069854736330,\n 0.13867411017417908,\n -0.07083784788846970,\n 0.53155958652496340,\n -0.008131757378578186,\n -0.20227709412574768,\n 0.08842383325099945,\n 0.86737281084060670,\n -0.29989498853683470,\n 0.09684963524341583,\n -0.031620755791664124,\n 0.58049958944320680,\n 0.11073039472103119,\n -0.25317376852035520,\n 0.56119495630264280,\n 0.34039902687072754,\n 1.21799755096435550,\n -0.021552175283432007,\n -0.15119931101799010,\n -0.054917752742767334,\n 0.80401933193206790,\n 0.010184556245803833,\n 0.06551001965999603,\n 0.19289986789226532,\n -0.21603244543075562,\n 0.025988996028900146,\n -0.65018039941787720,\n 0.50120383501052860,\n -0.26873117685317993,\n 0.19879785180091858,\n -0.26082211732864380,\n 0.46192330121994020,\n -0.55225759744644170,\n 0.01180829107761383,\n 0.05070953071117401,\n -0.67910110950469970,\n -0.04475890100002289,\n 0.36456465721130370,\n 0.13348321616649628,\n -0.01747957617044449,\n -0.13040468096733093,\n -0.16885089874267578,\n 0.33254384994506836,\n -0.29955965280532837,\n -0.26608777046203613,\n -0.10515909641981125,\n -0.32794988155364990,\n -0.05360160768032074,\n -0.21728354692459106,\n -0.39240026473999023,\n 0.33821657299995420,\n -0.008709043264389038,\n -0.43232405185699463,\n 0.63957405090332030,\n 0.35856172442436220,\n 0.45638042688369750,\n -0.13032492995262146,\n 0.46423870325088500,\n 0.29733401536941530,\n 0.21055299043655396,\n 0.33571115136146545,\n 0.27673989534378050,\n -1.99598908424377440,\n -0.43283250927925110,\n 0.27975231409072876,\n -0.27850526571273804,\n 0.48566606640815735,\n 0.14831756055355072,\n 0.12112566083669662,\n 0.37299168109893800,\n -0.36536252498626710,\n 0.51137733459472660,\n 0.26767280697822570,\n -0.35320448875427246,\n 0.13971213996410370,\n 0.01589038036763668,\n -0.30416059494018555,\n -0.35622701048851013,\n -0.29882031679153440,\n -0.0055990517139434814,\n -0.13705015182495117,\n 0.56266194581985470,\n -0.38767093420028687,\n -0.53967058658599850,\n 0.06725452840328217,\n -1.21143281459808350,\n 0.22121112048625946,\n 0.43655356764793396,\n 0.33551204204559326,\n -0.27671134471893310,\n -0.57174313068389890,\n -0.058216191828250885,\n 0.06912264972925186,\n -0.85613191127777100,\n 0.19724097847938538,\n 0.16659013926982880,\n 0.14675720036029816,\n 0.24287590384483337,\n -0.25835406780242920,\n 0.20815043151378632,\n -0.42670011520385740,\n 0.39319023489952090,\n 0.32358986139297485,\n 0.025813337415456772,\n -0.048157237470149994,\n -0.35826218128204346,\n 0.04717639088630676,\n 0.30715617537498474,\n -0.25429004430770874,\n -0.10547082126140594,\n 0.02966500073671341,\n 0.30983048677444460,\n -0.21249084174633026,\n -0.19436784088611603,\n 0.77192759513854980,\n -0.53250586986541750,\n 0.12887206673622130,\n 0.32633990049362180,\n -1.24625861644744870,\n -0.16218033432960510,\n -0.13100118935108185,\n -0.11232861876487732,\n 0.62874174118041990,\n -0.02767990529537201,\n -0.05274036154150963,\n -0.53563487529754640,\n -0.64365452527999880,\n -0.05103348195552826,\n 0.08929728716611862,\n -0.40029305219650270,\n -0.10031442344188690,\n 0.31065917015075684,\n -0.21383905410766602,\n 0.14335599541664124,\n -0.09037648141384125,\n -0.17714662849903107,\n 0.33824056386947630,\n -0.82754230499267580,\n 0.16252911090850830,\n -0.07532244920730591,\n -0.14527970552444458,\n -0.22860543429851532,\n -0.30257803201675415,\n 0.047085925936698914,\n 0.17316375672817230,\n 0.13065533339977264,\n -0.17600929737091064,\n 0.13690154254436493,\n -0.20214480161666870,\n 0.01934470236301422,\n 0.12377672642469406,\n -0.06723305583000183,\n 0.22328042984008790,\n -0.49700859189033510,\n -0.34658029675483704,\n 0.20576259493827820,\n -0.27207553386688230,\n -0.22085019946098328,\n -0.38435065746307373,\n 0.44076967239379883,\n 0.22373658418655396,\n -0.046322643756866455,\n 0.18001255393028260,\n 0.15259465575218200,\n 0.17097039520740510,\n -0.23130619525909424,\n 0.29231351613998413,\n -0.37276223301887510,\n -0.16314452886581420,\n 0.08766435086727142,\n -0.41216766834259033,\n -0.21740865707397460,\n -0.05558343976736069,\n -0.41705510020256040,\n -0.020109981298446655,\n -0.36541312932968140,\n -0.38398233056068420,\n 0.39583081007003784,\n 0.13429638743400574,\n -0.41582357883453370,\n 0.08929545432329178,\n 0.37230169773101807,\n -0.10493122041225433,\n -0.21501919627189636,\n -0.18377599120140076,\n 0.017690163105726242,\n -0.10214073956012726,\n 0.50946366786956790,\n -0.30789539217948914,\n -0.52317154407501220,\n -0.20865441858768463,\n 0.15374669432640076,\n 0.21395552158355713,\n -0.42342543601989746,\n 0.50638461112976070,\n 0.33827340602874756,\n 0.05084151029586792,\n 0.004318922758102417,\n 0.18277740478515625,\n 0.33151501417160034,\n -0.54885160923004150,\n 0.30941945314407350,\n 0.16642820835113525,\n -0.52071034908294680,\n -0.20474663376808167,\n -0.01535847783088684,\n 0.11818096041679382,\n ],\n \"clip_embeddings_magnitude\": 10.422953605651855,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[105, 144, 187]\",\n \"video_length\": None,\n \"in_trashcan\": False,\n \"removed\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"0\",\n \"size\": 6467070,\n \"subjectDistance\": 2.229,\n \"width\": 4032,\n \"main_file_id\": \"a5bcc802a708c07c18816d1b480a8b7d1\",\n },\n {\n \"thumbnail_big\": \"thumbnails_big/57dca8933514df1efafddf76004187cc1.webp\",\n \"square_thumbnail\": \"square_thumbnails/57dca8933514df1efafddf76004187cc1.webp\",\n \"square_thumbnail_small\": \"square_thumbnails_small/57dca8933514df1efafddf76004187cc1.webp\",\n \"added_on\": \"2023-06-16 16:30:56.866303 +00:00\",\n \"exif_gps_lat\": -33.9417083333333,\n \"exif_gps_lon\": 151.265258333333,\n \"exif_timestamp\": \"2017-08-17 10:32:47.000000 +00:00\",\n \"exif_json\": None,\n \"geolocation_json\": {\n \"type\": \"FeatureCollection\",\n \"query\": [151.265258, -33.941708],\n \"features\": [\n {\n \"id\": \"poi.721554610670\",\n \"text\": \"Mistral Point\",\n \"type\": \"Feature\",\n \"center\": [151.26523, -33.941658],\n \"context\": [\n {\n \"id\": \"postcode.503310\",\n \"text\": \"2035\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCNjRP\",\n },\n {\n \"id\": \"locality.269412878\",\n \"text\": \"Maroubra\",\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.26523, -33.941658],\n },\n \"relevance\": 1,\n \"place_name\": \"Mistral Point, Sydney, New South Wales 2035, Australia\",\n \"place_type\": [\"poi\"],\n \"properties\": {\n \"category\": \"historic site, historic\",\n \"landmark\": True,\n \"foursquare\": \"5958b8724420d86b1808f7bd\",\n },\n },\n {\n \"id\": \"postcode.503310\",\n \"bbox\": [151.205214, -33.958015, 151.265693, -33.931801],\n \"text\": \"2035\",\n \"type\": \"Feature\",\n \"center\": [151.242438, -33.942906],\n \"context\": [\n {\n \"id\": \"locality.269412878\",\n \"text\": \"Maroubra\",\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.242438, -33.942906],\n },\n \"relevance\": 1,\n \"place_name\": \"2035, Maroubra, New South Wales, Australia\",\n \"place_type\": [\"postcode\"],\n \"properties\": {\"mapbox_id\": \"dXJuOm1ieHBsYzpCNjRP\"},\n },\n {\n \"id\": \"locality.269412878\",\n \"bbox\": [151.226804557, -33.958002321, 151.292531935, -33.93283624],\n \"text\": \"Maroubra\",\n \"type\": \"Feature\",\n \"center\": [151.2575, -33.9475],\n \"context\": [\n {\n \"id\": \"place.24496142\",\n \"text\": \"Sydney\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [151.2575, -33.9475]},\n \"relevance\": 1,\n \"place_name\": \"Maroubra, New South Wales, Australia\",\n \"place_type\": [\"locality\"],\n \"properties\": {\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n },\n },\n {\n \"id\": \"place.24496142\",\n \"bbox\": [150.520934139, -34.11717528, 151.369884128, -33.562644328],\n \"text\": \"Sydney\",\n \"type\": \"Feature\",\n \"center\": [151.216454, -33.854816],\n \"context\": [\n {\n \"id\": \"region.33806\",\n \"text\": \"New South Wales\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.216454, -33.854816],\n },\n \"relevance\": 1,\n \"place_name\": \"Sydney, New South Wales, Australia\",\n \"place_type\": [\"place\"],\n \"properties\": {\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n },\n },\n {\n \"id\": \"region.33806\",\n \"bbox\": [140.999265, -37.5097258, 159.200456, -28.1370359],\n \"text\": \"New South Wales\",\n \"type\": \"Feature\",\n \"center\": [147.014694071448, -32.168971672412],\n \"context\": [\n {\n \"id\": \"country.8718\",\n \"text\": \"Australia\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n }\n ],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [147.014694071448, -32.168971672412],\n },\n \"relevance\": 1,\n \"place_name\": \"New South Wales, Australia\",\n \"place_type\": [\"region\"],\n \"properties\": {\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"short_code\": \"AU-NSW\",\n },\n },\n {\n \"id\": \"country.8718\",\n \"bbox\": [112.8256904, -54.8327658, 159.200456, -9.0436707],\n \"text\": \"Australia\",\n \"type\": \"Feature\",\n \"center\": [134.489562606981, -25.7349684916223],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [134.489562606981, -25.7349684916223],\n },\n \"relevance\": 1,\n \"place_name\": \"Australia\",\n \"place_type\": [\"country\"],\n \"properties\": {\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"short_code\": \"au\",\n },\n },\n ],\n \"attribution\": \"NOTICE: © 2023 Mapbox and its suppliers. All rights reserved. Use of this data is subject to the Mapbox Terms of Service (https://www.mapbox.com/about/maps/). This response and the information it contains may not be retained. POI(s) provided by Foursquare.\",\n \"search_text\": \"Mistral Point 2035 Maroubra Sydney New South Wales Australia\",\n },\n \"captions_json\": {\n \"places365\": {\n \"attributes\": [\n \"open area\",\n \"natural light\",\n \"far away horizon\",\n \"sunny\",\n \"rugged scene\",\n \"dirt\",\n \"natural\",\n \"dry\",\n \"clouds\",\n ],\n \"categories\": [],\n \"environment\": \"outdoor\",\n }\n },\n \"search_captions\": \"outdoor\",\n \"search_location\": \"Mistral Point 2035 Maroubra Sydney New South Wales Australia\",\n \"hidden\": False,\n \"public\": False,\n \"video\": False,\n \"clip_embeddings\": [\n -0.36666020750999450,\n -0.04985070228576660,\n -0.55112200975418090,\n -0.02282443828880787,\n 0.07433946430683136,\n -0.15077446401119232,\n -0.15087568759918213,\n 0.59861290454864500,\n 0.74118041992187500,\n 0.10699653625488281,\n 0.09809046983718872,\n -0.03829161822795868,\n 0.18715769052505493,\n -0.19874982535839080,\n -0.17219249904155730,\n -0.31639525294303894,\n -1.24799549579620360,\n -0.18328364193439484,\n 0.50648951530456540,\n 0.03295477479696274,\n 0.05792248249053955,\n -0.023687392473220825,\n 0.28321290016174316,\n -0.24516502022743225,\n -0.14424744248390198,\n 0.39068019390106200,\n 0.07994443178176880,\n -0.12718169391155243,\n 0.19193711876869202,\n -0.01991894096136093,\n -0.29576098918914795,\n 0.08526518195867538,\n -0.38953876495361330,\n -0.11270611733198166,\n -0.12309470027685165,\n -0.18142205476760864,\n -0.22653748095035553,\n 0.25507912039756775,\n 0.52982282638549800,\n -0.36807811260223390,\n -0.52154517173767090,\n -0.040144920349121094,\n 0.40785139799118040,\n -0.30712696909904480,\n 0.22090484201908112,\n -2.02335238456726070,\n 0.12097204476594925,\n 0.28021758794784546,\n 0.09183792769908905,\n 0.15624284744262695,\n 0.01185154914855957,\n 0.22275637090206146,\n 0.55167198181152340,\n -0.35499316453933716,\n 0.17891690135002136,\n 0.05891851335763931,\n 0.11313964426517487,\n -0.24982094764709473,\n -0.44442304968833923,\n -0.30212301015853880,\n -0.49696791172027590,\n -0.21229097247123718,\n -0.08969736099243164,\n 0.35271137952804565,\n -0.13295263051986694,\n -0.24579304456710815,\n -0.15918856859207153,\n 1.22507643699646000,\n -0.47429224848747253,\n -0.14982843399047852,\n -0.19515639543533325,\n -0.18184053897857666,\n -0.10420016944408417,\n -0.27105495333671570,\n -0.014386080205440521,\n -0.19864511489868164,\n -0.59518003463745120,\n 0.014985889196395874,\n -0.06511993706226349,\n -0.31187006831169130,\n 0.06655596941709518,\n 0.46725121140480040,\n -0.15730988979339600,\n 0.21003738045692444,\n 0.06053004041314125,\n 0.18053501844406128,\n 0.043053969740867615,\n -0.38710883259773254,\n -0.16783890128135680,\n -0.15172357857227325,\n 0.09899577498435974,\n -0.08344091475009918,\n -6.26640892028808600,\n 0.62121075391769410,\n 0.35546118021011350,\n 0.30253031849861145,\n -0.35551121830940247,\n -0.04457452893257141,\n -0.18457348644733430,\n -1.32966053485870360,\n 0.17469060420989990,\n -0.33178430795669556,\n 0.08215680718421936,\n 0.06592759490013123,\n -0.25799661874771120,\n 0.05711726099252701,\n -0.18006426095962524,\n -0.13938568532466888,\n 0.35842615365982056,\n -0.07131820172071457,\n 0.20586800575256348,\n -0.10146003961563110,\n -0.034379392862319946,\n 0.09676564484834671,\n -0.39100694656372070,\n -0.19553266465663910,\n -0.39932283759117126,\n 0.14101047813892365,\n 0.24242374300956726,\n -0.14971837401390076,\n 0.19587853550910950,\n -0.07761697471141815,\n -0.13720758259296417,\n -0.036738086491823196,\n 0.02775958925485611,\n 0.003398708999156952,\n 0.32153791189193726,\n -0.29471915960311890,\n -0.67542463541030880,\n 0.15028291940689087,\n -0.043420448899269104,\n -0.19056166708469390,\n 0.029275819659233093,\n 0.85994541645050050,\n -0.41359940171241760,\n -0.00037150830030441284,\n 0.21370775997638702,\n 0.10217175632715225,\n -0.15029510855674744,\n 0.15498718619346620,\n 0.14254488050937653,\n -0.12856674194335938,\n -0.12141317129135132,\n -0.21967242658138275,\n -0.22536158561706543,\n 0.40473008155822754,\n -0.29137128591537476,\n 1.18561625480651860,\n -0.30933171510696410,\n -0.04900578409433365,\n -0.22565384209156036,\n 0.026661649346351624,\n -0.53087574243545530,\n -0.02157217264175415,\n 0.61312061548233030,\n -0.68817430734634400,\n 0.55277901887893680,\n -0.02949278987944126,\n -0.35691511631011963,\n 0.17237737774848938,\n -0.84646338224411010,\n 0.28129857778549194,\n -0.001386452466249466,\n 0.13001629710197450,\n -0.34739860892295840,\n -0.10144264996051788,\n 0.64412295818328860,\n 0.38456526398658750,\n 0.23173999786376953,\n -0.17515747249126434,\n 0.020819609984755516,\n -0.32927608489990234,\n 0.08890204876661300,\n -0.15161202847957610,\n -0.18559980392456055,\n -0.04398374259471893,\n -0.27180469036102295,\n 0.11402454972267151,\n -0.30309695005416870,\n 0.71401089429855350,\n 0.58197832107543950,\n 0.09439059346914291,\n -0.11577179282903671,\n 0.33646234869956970,\n -0.22551852464675903,\n 0.13584175705909730,\n 0.21372529864311218,\n -0.58664178848266600,\n 0.24065378308296204,\n 0.34127143025398254,\n -0.17579154670238495,\n -0.33026531338691710,\n 0.15201087296009064,\n 0.36881905794143677,\n -0.50844454765319820,\n -0.48041981458663940,\n 0.003377307206392288,\n -0.21604734659194946,\n 1.22471618652343750,\n -0.34609639644622800,\n -0.07732010632753372,\n 0.19707258045673370,\n 0.03305445611476898,\n -0.25127333402633667,\n 0.02786330133676529,\n -0.13784091174602509,\n -0.33934032917022705,\n 0.11644668132066727,\n 0.12870812416076660,\n 0.28808489441871643,\n -0.060060858726501465,\n 0.05354182422161102,\n -0.18899542093276978,\n 0.08026752620935440,\n 0.14663285017013550,\n -0.10402667522430420,\n -0.04774125665426254,\n 0.78730720281600950,\n 0.14746253192424774,\n -0.08459271490573883,\n -0.32546862959861755,\n 0.32204133272171020,\n 0.08961439132690430,\n 0.05263507366180420,\n -0.65238183736801150,\n 0.03257568180561066,\n -0.52708566188812260,\n 0.38349610567092896,\n 0.15065722167491913,\n -0.003457099199295044,\n 0.05167556554079056,\n -0.36230301856994630,\n 0.13933974504470825,\n 0.31344616413116455,\n 0.32701274752616880,\n -1.06047749519348140,\n 0.03538133203983307,\n 0.014965444803237915,\n 0.14803899824619293,\n 0.08960707485675812,\n -0.25507515668869020,\n -0.26133170723915100,\n 0.22333431243896484,\n 0.38788953423500060,\n -0.26071867346763610,\n 0.34007954597473145,\n 0.05404023081064224,\n 0.24960926175117493,\n 0.35853916406631470,\n -0.24324321746826172,\n -0.25807657837867737,\n 0.15211161971092224,\n 0.03374246507883072,\n 0.09933222830295563,\n 0.14579382538795470,\n 0.09634265303611755,\n -0.15618667006492615,\n -0.020942021161317825,\n 1.40221035480499270,\n -0.02753184735774994,\n -0.018054869025945663,\n 0.15185260772705078,\n 0.36876207590103150,\n -0.07963977754116058,\n 0.21388293802738190,\n -0.57031232118606570,\n 0.16593083739280700,\n -0.12364573776721954,\n -0.19139841198921204,\n -0.07277653366327286,\n -0.07489527761936188,\n 0.06829735636711120,\n -0.09270213544368744,\n -0.58580970764160160,\n -0.57999855279922490,\n 0.10248814523220062,\n 0.07077595591545105,\n -0.11604766547679901,\n -0.10343541949987411,\n 0.09502948075532913,\n 0.07064349204301834,\n 0.31945759057998660,\n 0.26488274335861206,\n -0.11844316124916077,\n 0.10092554986476898,\n 0.05796369910240173,\n 0.007035255432128906,\n -0.20717875659465790,\n -0.39001202583312990,\n -0.23531873524188995,\n 0.029121950268745422,\n -0.31792175769805910,\n 0.21073228120803833,\n -0.25988331437110900,\n 0.01320233941078186,\n -0.04228278994560242,\n 0.87615823745727540,\n 0.23483057320117950,\n -0.14261351525783540,\n -0.03390306234359741,\n 0.05797864496707916,\n -0.08534366637468338,\n 0.60571086406707760,\n -0.40570425987243650,\n 0.39349102973937990,\n -0.0029862821102142334,\n 0.21820589900016785,\n 0.17360526323318481,\n 0.51648765802383420,\n 0.10162493586540222,\n -0.09098815917968750,\n 0.04716239869594574,\n 0.85775423049926760,\n -0.11656519025564194,\n 0.23683629930019380,\n 0.07459381222724915,\n 0.55665183067321780,\n -0.03265659511089325,\n -0.03649625927209854,\n 0.85287141799926760,\n 0.35186266899108887,\n 0.036931782960891724,\n -0.26376897096633910,\n -0.45744919776916504,\n -0.09645552933216095,\n 0.66262829303741460,\n -0.10217510163784027,\n 0.08431930840015411,\n -0.050179481506347656,\n -0.12746801972389220,\n -0.37214517593383790,\n -0.06164243817329407,\n 0.26156315207481384,\n -0.14883297681808472,\n 0.16589079797267914,\n -0.35074496269226074,\n 0.08279553055763245,\n -0.13153877854347230,\n 0.16450478136539460,\n -0.13764882087707520,\n -0.20287595689296722,\n -0.046635668724775314,\n 0.12855112552642822,\n 0.24670921266078950,\n -0.08000148832798004,\n -0.09557560086250305,\n -0.25575160980224610,\n 0.15930658578872680,\n -0.67159068584442140,\n -0.07150696963071823,\n 0.17778603732585907,\n 0.25455826520919800,\n -0.23019903898239136,\n -0.08271875232458115,\n -0.10421910881996155,\n 0.55327570438385010,\n 0.17031423747539520,\n -0.99103516340255740,\n 0.73087739944458010,\n 0.33844703435897827,\n 0.61791002750396730,\n -0.018104268237948418,\n 0.42206788063049316,\n -0.07971519231796265,\n 0.06754805147647858,\n 0.18134978413581848,\n 0.07568907737731934,\n -1.62685477733612060,\n -0.38066196441650390,\n 0.66141670942306520,\n -0.25981235504150390,\n 0.55625867843627930,\n 0.51772624254226680,\n 0.24509307742118835,\n -0.20850536227226257,\n -0.16368716955184937,\n 0.76303339004516600,\n -0.0018448233604431152,\n -0.05956418812274933,\n -0.13241642713546753,\n 0.005002804100513458,\n -0.28848242759704590,\n -0.44567650556564330,\n -0.09337581694126129,\n 0.34333360195159910,\n 0.18656188249588013,\n 0.39478647708892820,\n 0.22161819040775300,\n -0.41303259134292600,\n -0.42747217416763306,\n -1.40258526802062990,\n 0.58988147974014280,\n 0.41526830196380615,\n 0.22585415840148926,\n -0.047290802001953125,\n -0.40906193852424620,\n 0.13364250957965850,\n 0.36895215511322020,\n -0.32853290438652040,\n 0.018866248428821564,\n 0.021997720003128052,\n 0.33384290337562560,\n 0.12766578793525696,\n -0.25208842754364014,\n 0.29037922620773315,\n -0.26148244738578796,\n 0.03055279701948166,\n -0.048704586923122406,\n -0.019998198375105858,\n -0.22977681457996368,\n 0.18012699484825134,\n 0.05622486397624016,\n 0.22875103354454040,\n -0.09111046046018600,\n -0.21855254471302032,\n -0.021557403728365898,\n -0.14998503029346466,\n -0.33238950371742250,\n -0.44446778297424316,\n 0.64051526784896850,\n -0.13707451522350310,\n 0.11501492559909820,\n 0.30077239871025085,\n -1.04448294639587400,\n -0.75157833099365230,\n -0.31508189439773560,\n -0.18117314577102660,\n 0.04961024224758148,\n 0.020732292905449867,\n 0.51410740613937380,\n -0.44715842604637146,\n -0.59480583667755130,\n 0.25802063941955566,\n -0.005115009844303131,\n -0.66136705875396730,\n -0.37843626737594604,\n 0.39065134525299070,\n -0.16299697756767273,\n -0.08189993351697922,\n 0.07605299353599548,\n 0.15921586751937866,\n 0.55074018239974980,\n -0.66295832395553590,\n 0.12967866659164430,\n 0.17624089121818542,\n -0.23016670346260070,\n -0.24796481430530548,\n -0.45518487691879270,\n -0.11801701784133911,\n 0.07801371812820435,\n 0.02105349302291870,\n -0.15974965691566467,\n -0.039424389600753784,\n -0.56841480731964110,\n -0.24004876613616943,\n 0.19917711615562440,\n 0.05348219349980354,\n -0.09810043871402740,\n -0.06265527009963989,\n -0.0035811271518468857,\n 0.44162738323211670,\n -0.20175644755363464,\n 0.08200869709253311,\n -0.12142806500196457,\n 0.46088671684265137,\n 0.12649869918823242,\n -0.39095637202262880,\n 0.16196821630001068,\n 0.42821982502937317,\n 0.53625297546386720,\n -0.10519468784332275,\n 0.06364251673221588,\n -0.43340182304382324,\n -0.27793857455253600,\n 0.07457497715950012,\n -0.17316223680973053,\n -0.13105086982250214,\n -0.11595485359430313,\n -0.43414521217346190,\n 0.013297945261001587,\n -0.03564862906932831,\n -0.04748070240020752,\n 0.43650120496749880,\n 0.06940247118473053,\n -0.04091569781303406,\n -0.42648598551750183,\n 0.44560009241104126,\n -0.27538335323333740,\n -0.24719163775444030,\n -0.39051297307014465,\n 0.10344080626964569,\n -0.05663326010107994,\n 0.26151871681213380,\n -0.47625967860221863,\n 0.16123828291893005,\n -0.36013549566268920,\n 0.011993957683444023,\n 0.20137929916381836,\n -0.14949794113636017,\n 0.36271947622299194,\n 0.36080580949783325,\n 0.16761866211891174,\n 0.12525813281536102,\n 0.41298842430114746,\n -0.17537128925323486,\n -0.43780979514122010,\n 0.31959301233291626,\n 0.56603193283081050,\n -1.01032650470733640,\n -0.23371657729148865,\n -0.31553673744201660,\n -0.010249286890029907,\n ],\n \"clip_embeddings_magnitude\": 10.27176284790039,\n \"aspect_ratio\": 1.33,\n \"rating\": 0,\n \"dominant_color\": \"[152, 188, 233]\",\n \"video_length\": None,\n \"in_trashcan\": False,\n \"removed\": False,\n \"timestamp\": None,\n \"camera\": \"Pixel 2\",\n \"digitalZoomRatio\": None,\n \"focalLength35Equivalent\": None,\n \"focal_length\": 4.442,\n \"fstop\": 1.8,\n \"height\": 3024,\n \"iso\": None,\n \"lens\": None,\n \"shutter_speed\": \"0\",\n \"size\": 8068575,\n \"subjectDistance\": 3.289,\n \"width\": 4032,\n \"main_file_id\": \"57dca8933514df1efafddf76004187cc1\",\n },\n]\n" }, { "path": "api/tests/fixtures/api_util/sunburst_expectation.py", "content": "expectation = {\n \"name\": \"Places I've visited\",\n \"children\": [\n {\n \"name\": \"Australia\",\n \"children\": [\n {\n \"name\": \"New South Wales\",\n \"children\": [{\"name\": \"Sydney\", \"value\": 4, \"hex\": \"#57d3db\"}],\n \"hex\": \"#b9db57\",\n }\n ],\n \"hex\": \"#57d3db\",\n },\n {\n \"name\": \"Canada\",\n \"children\": [\n {\n \"name\": \"Ontario\",\n \"children\": [\n {\"name\": \"Peterborough County\", \"value\": 1, \"hex\": \"#c957db\"}\n ],\n \"hex\": \"#dbae57\",\n }\n ],\n \"hex\": \"#dbae57\",\n },\n {\n \"name\": \"Germany\",\n \"children\": [\n {\n \"name\": \"Berlin\",\n \"children\": [\n {\"name\": \"Friedrichshain\", \"value\": 1, \"hex\": \"#db5f57\"},\n {\"name\": \"Kreuzberg\", \"value\": 1, \"hex\": \"#69db57\"},\n ],\n \"hex\": \"#db579e\",\n }\n ],\n \"hex\": \"#57d3db\",\n },\n {\n \"name\": \"India\",\n \"children\": [\n {\n \"name\": \"Ladakh\",\n \"children\": [{\"name\": \"Leh\", \"value\": 2, \"hex\": \"#c957db\"}],\n \"hex\": \"#5784db\",\n }\n ],\n \"hex\": \"#db579e\",\n },\n ],\n}\n" }, { "path": "api/tests/fixtures/geocode/__init__.py", "content": "" }, { "path": "api/tests/fixtures/geocode/expectations/mapbox.py", "content": "from api.geocode import GEOCODE_VERSION\n\nexpectations = [\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Beach Road\", \"center\": [-33.888012425, 151.275216500055]},\n {\"text\": \"Bondi Beach\", \"center\": [-33.888012425, 151.275216500055]},\n {\"text\": \"Sydney\", \"center\": [-33.888012425, 151.275216500055]},\n {\"text\": \"New South Wales\", \"center\": [-33.888012425, 151.275216500055]},\n {\"text\": \"Australia\", \"center\": [-33.888012425, 151.275216500055]},\n ],\n \"places\": [\n \"Beach Road\",\n \"Bondi Beach\",\n \"Sydney\",\n \"New South Wales\",\n \"Australia\",\n ],\n \"address\": \"17 Beach Road, Bondi Beach New South Wales 2026, Australia\",\n \"center\": [-33.888012425, 151.275216500055],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Fire Route 47\", \"center\": [44.5540639, -78.1955566]},\n {\"text\": \"Lakefield\", \"center\": [44.5540639, -78.1955566]},\n {\"text\": \"Peterborough County\", \"center\": [44.5540639, -78.1955566]},\n {\"text\": \"Ontario\", \"center\": [44.5540639, -78.1955566]},\n {\"text\": \"Canada\", \"center\": [44.5540639, -78.1955566]},\n ],\n \"places\": [\n \"Fire Route 47\",\n \"Lakefield\",\n \"Peterborough County\",\n \"Ontario\",\n \"Canada\",\n ],\n \"address\": \"2880 Fire Route 47, Lakefield, Ontario K0L 2H0, Canada\",\n \"center\": [44.5540639, -78.1955566],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Roncalli\", \"center\": [52.501711, 13.381502]},\n {\"text\": \"Kreuzberg\", \"center\": [52.501711, 13.381502]},\n {\"text\": \"Berlin\", \"center\": [52.501711, 13.381502]},\n {\"text\": \"Germany\", \"center\": [52.501711, 13.381502]},\n ],\n \"places\": [\"Roncalli\", \"Kreuzberg\", \"Berlin\", \"Germany\"],\n \"address\": \"Roncalli, Mckernstr, Berlin, 10963, Germany\",\n \"center\": [52.501711, 13.381502],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Lakeshore Road\", \"center\": [33.913368523661, 78.45710763764029]},\n {\"text\": \"Shachokol\", \"center\": [33.913368523661, 78.45710763764029]},\n {\"text\": \"Leh\", \"center\": [33.913368523661, 78.45710763764029]},\n {\"text\": \"Leh\", \"center\": [33.913368523661, 78.45710763764029]},\n {\"text\": \"Ladakh\", \"center\": [33.913368523661, 78.45710763764029]},\n {\"text\": \"India\", \"center\": [33.913368523661, 78.45710763764029]},\n ],\n \"places\": [\n \"Lakeshore Road\",\n \"Shachokol\",\n \"Leh\",\n \"Leh\",\n \"Ladakh\",\n \"India\",\n ],\n \"address\": \"Lakeshore Road ، 194201 Leh، India\",\n \"center\": [33.913368523661, 78.45710763764029],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Mahon Rock Pool\", \"center\": [-33.94292, 151.263303]},\n {\"text\": \"Maroubra\", \"center\": [-33.94292, 151.263303]},\n {\"text\": \"Sydney\", \"center\": [-33.94292, 151.263303]},\n {\"text\": \"New South Wales\", \"center\": [-33.94292, 151.263303]},\n {\"text\": \"Australia\", \"center\": [-33.94292, 151.263303]},\n ],\n \"places\": [\n \"Mahon Rock Pool\",\n \"Maroubra\",\n \"Sydney\",\n \"New South Wales\",\n \"Australia\",\n ],\n \"address\": \"Mahon Rock Pool, Marine Pde., Sydney, New South Wales, Australia\",\n \"center\": [-33.94292, 151.263303],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Main Bazaar\", \"center\": [34.1622089, 77.585527]},\n {\"text\": \"Chuchat Yakma\", \"center\": [34.1622089, 77.585527]},\n {\"text\": \"Leh\", \"center\": [34.1622089, 77.585527]},\n {\"text\": \"Leh\", \"center\": [34.1622089, 77.585527]},\n {\"text\": \"Ladakh\", \"center\": [34.1622089, 77.585527]},\n {\"text\": \"India\", \"center\": [34.1622089, 77.585527]},\n ],\n \"places\": [\n \"Main Bazaar\",\n \"Chuchat Yakma\",\n \"Leh\",\n \"Leh\",\n \"Ladakh\",\n \"India\",\n ],\n \"address\": \"Main Bazaar ، 194101 Leh، India\",\n \"center\": [34.1622089, 77.585527],\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/expectations/nominatim.py", "content": "from api.geocode import GEOCODE_VERSION\n\nexpectations = [\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Beach Road\", \"center\": [-33.88801645, 151.27521180010973]},\n {\"text\": \"Seven Ways\", \"center\": [-33.88801645, 151.27521180010973]},\n {\"text\": \"Bondi Beach\", \"center\": [-33.88801645, 151.27521180010973]},\n {\"text\": \"Eastern Suburbs\", \"center\": [-33.88801645, 151.27521180010973]},\n {\"text\": \"Sydney\", \"center\": [-33.88801645, 151.27521180010973]},\n {\"text\": \"New South Wales\", \"center\": [-33.88801645, 151.27521180010973]},\n {\"text\": \"Australia\", \"center\": [-33.88801645, 151.27521180010973]},\n ],\n \"places\": [\n \"Beach Road\",\n \"Seven Ways\",\n \"Bondi Beach\",\n \"Eastern Suburbs\",\n \"Sydney\",\n \"New South Wales\",\n \"Australia\",\n ],\n \"address\": \"17, Beach Road, Seven Ways, Bondi Beach, Eastern Suburbs, Sydney, Waverley Council, New South Wales, 2026, Australia\",\n \"center\": [-33.88801645, 151.27521180010973],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\n \"text\": \"Fire Route 47\",\n \"center\": [44.55373905094425, -78.19571323702382],\n },\n {\"text\": \"Selwyn\", \"center\": [44.55373905094425, -78.19571323702382]},\n {\n \"text\": \"Peterborough County\",\n \"center\": [44.55373905094425, -78.19571323702382],\n },\n {\"text\": \"Ontario\", \"center\": [44.55373905094425, -78.19571323702382]},\n {\"text\": \"Canada\", \"center\": [44.55373905094425, -78.19571323702382]},\n ],\n \"places\": [\n \"Fire Route 47\",\n \"Selwyn\",\n \"Peterborough County\",\n \"Ontario\",\n \"Canada\",\n ],\n \"address\": \"3118, Fire Route 47, Selwyn, Peterborough County, Central Ontario, Ontario, K0L 2C0, Canada\",\n \"center\": [44.55373905094425, -78.19571323702382],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\n \"text\": \"Möckernstraße\",\n \"center\": [52.501606300000006, 13.381190860617572],\n },\n {\"text\": \"Kreuzberg\", \"center\": [52.501606300000006, 13.381190860617572]},\n {\n \"text\": \"Friedrichshain-Kreuzberg\",\n \"center\": [52.501606300000006, 13.381190860617572],\n },\n {\"text\": \"Berlin\", \"center\": [52.501606300000006, 13.381190860617572]},\n {\"text\": \"Deutschland\", \"center\": [52.501606300000006, 13.381190860617572]},\n ],\n \"places\": [\n \"Möckernstraße\",\n \"Kreuzberg\",\n \"Friedrichshain-Kreuzberg\",\n \"Berlin\",\n \"Deutschland\",\n ],\n \"address\": \"Tempodrom, 10, Möckernstraße, Kreuzberg, Friedrichshain-Kreuzberg, Berlin, 10963, Deutschland\",\n \"center\": [52.501606300000006, 13.381190860617572],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Lakeshore road\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Spangmik\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Leh Tehsil\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Ladakh\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"India\", \"center\": [33.9132578, 78.4571752]},\n ],\n \"places\": [\n \"Lakeshore road\",\n \"Spangmik\",\n \"Leh Tehsil\",\n \"Ladakh\",\n \"India\",\n ],\n \"address\": \"Camp Water Mark, Lakeshore road, Spangmik, Leh Tehsil, Leh District, Ladakh, India\",\n \"center\": [33.9132578, 78.4571752],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Marine Parade\", \"center\": [-33.9430026, 151.26386704076833]},\n {\"text\": \"Maroubra\", \"center\": [-33.9430026, 151.26386704076833]},\n {\"text\": \"Eastern Suburbs\", \"center\": [-33.9430026, 151.26386704076833]},\n {\"text\": \"Sydney\", \"center\": [-33.9430026, 151.26386704076833]},\n {\"text\": \"New South Wales\", \"center\": [-33.9430026, 151.26386704076833]},\n {\"text\": \"Australia\", \"center\": [-33.9430026, 151.26386704076833]},\n ],\n \"places\": [\n \"Marine Parade\",\n \"Maroubra\",\n \"Eastern Suburbs\",\n \"Sydney\",\n \"New South Wales\",\n \"Australia\",\n ],\n \"address\": \"Mahon Pool, Marine Parade, Maroubra, Eastern Suburbs, Sydney, Randwick City Council, New South Wales, 2035, Australia\",\n \"center\": [-33.9430026, 151.26386704076833],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Main Bazaar\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"Leh\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"Leh Tehsil\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"Ladakh\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"India\", \"center\": [34.1621176, 77.585862]},\n ],\n \"places\": [\"Main Bazaar\", \"Leh\", \"Leh Tehsil\", \"Ladakh\", \"India\"],\n \"address\": \"Dry Fruit Market, Main Bazaar, Matsik Chulung, Leh, Leh Tehsil, Leh District, Ladakh, India\",\n \"center\": [34.1621176, 77.585862],\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/expectations/opencage.py", "content": "from api.geocode import GEOCODE_VERSION\n\nexpectations = [\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Beach Road\", \"center\": [-33.8880165, 151.2752118]},\n {\"text\": \"Bondi Beach\", \"center\": [-33.8880165, 151.2752118]},\n {\"text\": \"Waverley Council\", \"center\": [-33.8880165, 151.2752118]},\n {\"text\": \"Eastern Suburbs\", \"center\": [-33.8880165, 151.2752118]},\n {\"text\": \"New South Wales\", \"center\": [-33.8880165, 151.2752118]},\n {\"text\": \"Australia\", \"center\": [-33.8880165, 151.2752118]},\n ],\n \"places\": [\n \"Beach Road\",\n \"Bondi Beach\",\n \"Waverley Council\",\n \"Eastern Suburbs\",\n \"New South Wales\",\n \"Australia\",\n ],\n \"address\": \"17 Beach Road, Bondi Beach NSW 2026, Australia\",\n \"center\": [-33.8880165, 151.2752118],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Fire Route 47\", \"center\": [44.5537391, -78.1957132]},\n {\"text\": \"Ontario\", \"center\": [44.5537391, -78.1957132]},\n {\"text\": \"Peterborough County\", \"center\": [44.5537391, -78.1957132]},\n {\"text\": \"Canada\", \"center\": [44.5537391, -78.1957132]},\n ],\n \"places\": [\"Fire Route 47\", \"Ontario\", \"Peterborough County\", \"Canada\"],\n \"address\": \"3118 Fire Route 47, Selwyn, ON K0L 2C0, Canada\",\n \"center\": [44.5537391, -78.1957132],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Tempodrom\", \"center\": [52.5016063, 13.3811909]},\n {\"text\": \"Möckernstraße\", \"center\": [52.5016063, 13.3811909]},\n {\"text\": \"Kreuzberg\", \"center\": [52.5016063, 13.3811909]},\n {\"text\": \"Friedrichshain-Kreuzberg\", \"center\": [52.5016063, 13.3811909]},\n {\"text\": \"Berlin\", \"center\": [52.5016063, 13.3811909]},\n {\"text\": \"Germany\", \"center\": [52.5016063, 13.3811909]},\n ],\n \"places\": [\n \"Tempodrom\",\n \"Möckernstraße\",\n \"Kreuzberg\",\n \"Friedrichshain-Kreuzberg\",\n \"Berlin\",\n \"Germany\",\n ],\n \"address\": \"Tempodrom, Möckernstraße 10, 10963 Berlin, Germany\",\n \"center\": [52.5016063, 13.3811909],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Camp Water Mark\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Lakeshore road\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Spangmik\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Ladakh\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"Leh Tehsil\", \"center\": [33.9132578, 78.4571752]},\n {\"text\": \"India\", \"center\": [33.9132578, 78.4571752]},\n ],\n \"places\": [\n \"Camp Water Mark\",\n \"Lakeshore road\",\n \"Spangmik\",\n \"Ladakh\",\n \"Leh Tehsil\",\n \"India\",\n ],\n \"address\": \"Camp Water Mark, Lakeshore road, Leh district, Spangmik -, Ladakh, India\",\n \"center\": [33.9132578, 78.4571752],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Mahon Pool\", \"center\": [-33.9430026, 151.263867]},\n {\"text\": \"Marine Parade\", \"center\": [-33.9430026, 151.263867]},\n {\"text\": \"Maroubra\", \"center\": [-33.9430026, 151.263867]},\n {\"text\": \"Randwick City Council\", \"center\": [-33.9430026, 151.263867]},\n {\"text\": \"Eastern Suburbs\", \"center\": [-33.9430026, 151.263867]},\n {\"text\": \"New South Wales\", \"center\": [-33.9430026, 151.263867]},\n {\"text\": \"Australia\", \"center\": [-33.9430026, 151.263867]},\n ],\n \"places\": [\n \"Mahon Pool\",\n \"Marine Parade\",\n \"Maroubra\",\n \"Randwick City Council\",\n \"Eastern Suburbs\",\n \"New South Wales\",\n \"Australia\",\n ],\n \"address\": \"Mahon Pool, Marine Parade, Maroubra NSW 2035, Australia\",\n \"center\": [-33.9430026, 151.263867],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Dry Fruit Market\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"Main Bazaar\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"Ladakh\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"Leh Tehsil\", \"center\": [34.1621176, 77.585862]},\n {\"text\": \"India\", \"center\": [34.1621176, 77.585862]},\n ],\n \"places\": [\n \"Dry Fruit Market\",\n \"Main Bazaar\",\n \"Ladakh\",\n \"Leh Tehsil\",\n \"India\",\n ],\n \"address\": \"Dry Fruit Market, Main Bazaar, Leh district, Leh -, Ladakh, India\",\n \"center\": [34.1621176, 77.585862],\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/expectations/tomtom.py", "content": "from api.geocode import GEOCODE_VERSION\n\nexpectations = [\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Beach Road\", \"center\": [-33.888145, 151.275085]},\n {\"text\": \"Bondi Beach\", \"center\": [-33.888145, 151.275085]},\n {\"text\": \"New South Wales\", \"center\": [-33.888145, 151.275085]},\n {\"text\": \"Sydney\", \"center\": [-33.888145, 151.275085]},\n {\"text\": \"Australia\", \"center\": [-33.888145, 151.275085]},\n ],\n \"places\": [\n \"Beach Road\",\n \"Bondi Beach\",\n \"New South Wales\",\n \"Sydney\",\n \"Australia\",\n ],\n \"address\": \"17 Beach Road, Bondi Beach, New South Wales, 2026\",\n \"center\": [-33.888145, 151.275085],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Fire Route 47\", \"center\": [44.553463, -78.195114]},\n {\"text\": \"Lakefield\", \"center\": [44.553463, -78.195114]},\n {\"text\": \"Canada\", \"center\": [44.553463, -78.195114]},\n ],\n \"places\": [\"Fire Route 47\", \"Lakefield\", \"Canada\"],\n \"address\": \"2876 Fire Route 47, Lakefield ON K0L 2H0\",\n \"center\": [44.553463, -78.195114],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Möckernstraße\", \"center\": [52.501957, 13.382298]},\n {\"text\": \"Kreuzberg\", \"center\": [52.501957, 13.382298]},\n {\"text\": \"Berlin\", \"center\": [52.501957, 13.382298]},\n {\"text\": \"Deutschland\", \"center\": [52.501957, 13.382298]},\n ],\n \"places\": [\"Möckernstraße\", \"Kreuzberg\", \"Berlin\", \"Deutschland\"],\n \"address\": \"Möckernstraße 138, 10963 Berlin\",\n \"center\": [52.501957, 13.382298],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Thiksey\", \"center\": [33.913391, 78.457077]},\n {\"text\": \"Ladakh\", \"center\": [33.913391, 78.457077]},\n {\"text\": \"Leh\", \"center\": [33.913391, 78.457077]},\n {\"text\": \"India\", \"center\": [33.913391, 78.457077]},\n ],\n \"places\": [\"Thiksey\", \"Ladakh\", \"Leh\", \"India\"],\n \"address\": \"Thiksey, Ladakh 194101, Ladakh\",\n \"center\": [33.913391, 78.457077],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Marine Parade\", \"center\": [-33.943535, 151.26181]},\n {\"text\": \"Maroubra\", \"center\": [-33.943535, 151.26181]},\n {\"text\": \"New South Wales\", \"center\": [-33.943535, 151.26181]},\n {\"text\": \"Sydney\", \"center\": [-33.943535, 151.26181]},\n {\"text\": \"Australia\", \"center\": [-33.943535, 151.26181]},\n ],\n \"places\": [\n \"Marine Parade\",\n \"Maroubra\",\n \"New South Wales\",\n \"Sydney\",\n \"Australia\",\n ],\n \"address\": \"106 Marine Parade, Maroubra, New South Wales, 2035\",\n \"center\": [-33.943535, 151.26181],\n },\n {\n \"_v\": GEOCODE_VERSION,\n \"features\": [\n {\"text\": \"Leh Ladakh\", \"center\": [34.16209, 77.585808]},\n {\"text\": \"Ladakh\", \"center\": [34.16209, 77.585808]},\n {\"text\": \"Leh\", \"center\": [34.16209, 77.585808]},\n {\"text\": \"India\", \"center\": [34.16209, 77.585808]},\n ],\n \"places\": [\"Leh Ladakh\", \"Ladakh\", \"Leh\", \"India\"],\n \"address\": \"Leh Ladakh, Ladakh 194101, Ladakh\",\n \"center\": [34.16209, 77.585808],\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/responses/mapbox.py", "content": "responses = [\n {\n \"id\": \"address.2007486836507642\",\n \"type\": \"Feature\",\n \"place_type\": [\"address\"],\n \"relevance\": 1,\n \"properties\": {\n \"accuracy\": \"rooftop\",\n \"mapbox_id\": \"dXJuOm1ieGFkcjo3NmQ1MTQzZi1kZmEwLTQ4NzAtYTAyNy0zYWY1MGJiZTIxYWU\",\n },\n \"text\": \"Beach Road\",\n \"place_name\": \"17 Beach Road, Bondi Beach New South Wales 2026, Australia\",\n \"center\": [151.275216500055, -33.888012425],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [151.275216500055, -33.888012425],\n },\n \"address\": \"17\",\n \"context\": [\n {\n \"id\": \"postcode.429582\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCbzRP\",\n \"text\": \"2026\",\n },\n {\n \"id\": \"locality.259156494\",\n \"wikidata\": \"Q673418\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpEM0pxRGc\",\n \"text\": \"Bondi Beach\",\n },\n {\n \"id\": \"place.24496142\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n \"text\": \"Sydney\",\n },\n {\n \"id\": \"region.33806\",\n \"short_code\": \"AU-NSW\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"text\": \"New South Wales\",\n },\n {\n \"id\": \"country.8718\",\n \"short_code\": \"au\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"text\": \"Australia\",\n },\n ],\n },\n {\n \"id\": \"address.1654238032706288\",\n \"type\": \"Feature\",\n \"place_type\": [\"address\"],\n \"relevance\": 1,\n \"properties\": {\n \"accuracy\": \"point\",\n \"mapbox_id\": \"dXJuOm1ieGFkcjoyOTlkZDc4Yy0wYWVlLTQ2ZGUtYmFkOC04ZDA0Y2VhYzNkZGQ\",\n },\n \"text\": \"Fire Route 47\",\n \"place_name\": \"2880 Fire Route 47, Lakefield, Ontario K0L 2H0, Canada\",\n \"center\": [-78.1955566, 44.5540639],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [-78.1955566, 44.5540639]},\n \"address\": \"2880\",\n \"context\": [\n {\n \"id\": \"postcode.2669825575\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpueUpPSnc\",\n \"text\": \"K0L 2H0\",\n },\n {\n \"id\": \"place.106145831\",\n \"wikidata\": \"Q114506323\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpCbE9vSnc\",\n \"text\": \"Lakefield\",\n },\n {\n \"id\": \"district.1140263\",\n \"wikidata\": \"Q730542\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFV1lu\",\n \"text\": \"Peterborough County\",\n },\n {\n \"id\": \"region.17447\",\n \"short_code\": \"CA-ON\",\n \"wikidata\": \"Q1904\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpSQ2M\",\n \"text\": \"Ontario\",\n },\n {\n \"id\": \"country.8743\",\n \"short_code\": \"ca\",\n \"wikidata\": \"Q16\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJaWM\",\n \"text\": \"Canada\",\n },\n ],\n },\n {\n \"id\": \"poi.1047972024770\",\n \"type\": \"Feature\",\n \"place_type\": [\"poi\"],\n \"relevance\": 1,\n \"properties\": {\n \"foursquare\": \"50d22a90e4b0d3035504ae2b\",\n \"landmark\": True,\n \"address\": \"Möckernstr\",\n \"category\": \"circus\",\n \"maki\": \"theatre\",\n },\n \"text\": \"Roncalli\",\n \"place_name\": \"Roncalli, Mckernstr, Berlin, 10963, Germany\",\n \"center\": [13.381502, 52.501711],\n \"geometry\": {\"coordinates\": [13.381502, 52.501711], \"type\": \"Point\"},\n \"context\": [\n {\n \"id\": \"postcode.5541434\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpWSTQ2\",\n \"text\": \"10963\",\n },\n {\n \"id\": \"locality.181160506\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpDc3hLT2c\",\n \"text\": \"Kreuzberg\",\n },\n {\n \"id\": \"place.115770\",\n \"short_code\": \"DE-BE\",\n \"wikidata\": \"Q64\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBY1E2\",\n \"text\": \"Berlin\",\n },\n {\n \"id\": \"country.8762\",\n \"short_code\": \"de\",\n \"wikidata\": \"Q183\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJam8\",\n \"text\": \"Germany\",\n },\n ],\n },\n {\n \"id\": \"address.4021415981422040\",\n \"type\": \"Feature\",\n \"place_type\": [\"address\"],\n \"relevance\": 1,\n \"properties\": {\"accuracy\": \"street\"},\n \"text\": \"Lakeshore Road\",\n \"place_name\": \"Lakeshore Road ، 194201 Leh، India\",\n \"center\": [78.45710763764029, 33.913368523661],\n \"geometry\": {\n \"type\": \"Point\",\n \"coordinates\": [78.45710763764029, 33.913368523661],\n },\n \"context\": [\n {\n \"id\": \"postcode.13119083\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzp5QzVy\",\n \"text\": \"194201\",\n },\n {\n \"id\": \"locality.3711994475\",\n \"wikidata\": \"Q24912826\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzozVUNLYXc\",\n \"text\": \"Shachokol\",\n },\n {\n \"id\": \"place.25135211\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n \"text\": \"Leh\",\n },\n {\n \"id\": \"district.3196523\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n \"text\": \"Leh\",\n },\n {\n \"id\": \"region.222315\",\n \"short_code\": \"IN-LA\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"text\": \"Ladakh\",\n },\n {\n \"id\": \"country.8811\",\n \"short_code\": \"in\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"text\": \"India\",\n },\n ],\n },\n {\n \"id\": \"poi.695784709457\",\n \"type\": \"Feature\",\n \"place_type\": [\"poi\"],\n \"relevance\": 1,\n \"properties\": {\n \"foursquare\": \"4b550d36f964a52055d927e3\",\n \"landmark\": True,\n \"address\": \"Marine Pde.\",\n \"category\": \"swimming pool, pool, swim club\",\n },\n \"text\": \"Mahon Rock Pool\",\n \"place_name\": \"Mahon Rock Pool, Marine Pde., Sydney, New South Wales, Australia\",\n \"center\": [151.263303, -33.94292],\n \"geometry\": {\"coordinates\": [151.263303, -33.94292], \"type\": \"Point\"},\n \"context\": [\n {\n \"id\": \"locality.269412878\",\n \"wikidata\": \"Q2914843\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpFQTdxRGc\",\n \"text\": \"Maroubra\",\n },\n {\n \"id\": \"place.24496142\",\n \"wikidata\": \"Q3130\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWFhJRGc\",\n \"text\": \"Sydney\",\n },\n {\n \"id\": \"region.33806\",\n \"short_code\": \"AU-NSW\",\n \"wikidata\": \"Q3224\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpoQTQ\",\n \"text\": \"New South Wales\",\n },\n {\n \"id\": \"country.8718\",\n \"short_code\": \"au\",\n \"wikidata\": \"Q408\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJZzQ\",\n \"text\": \"Australia\",\n },\n ],\n },\n {\n \"id\": \"address.15438142488176\",\n \"type\": \"Feature\",\n \"place_type\": [\"address\"],\n \"relevance\": 1,\n \"properties\": {\"accuracy\": \"street\"},\n \"text\": \"Main Bazaar\",\n \"place_name\": \"Main Bazaar ، 194101 Leh، India\",\n \"center\": [77.585527, 34.1622089],\n \"geometry\": {\"type\": \"Point\", \"coordinates\": [77.585527, 34.1622089]},\n \"context\": [\n {\n \"id\": \"postcode.13053547\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzp4eTVy\",\n \"text\": \"194101\",\n },\n {\n \"id\": \"locality.997100139\",\n \"wikidata\": \"Q24909733\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpPMjZLYXc\",\n \"text\": \"Chuchat Yakma\",\n },\n {\n \"id\": \"place.25135211\",\n \"wikidata\": \"Q230818\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBWCtJYXc\",\n \"text\": \"Leh\",\n },\n {\n \"id\": \"district.3196523\",\n \"wikidata\": \"Q1921210\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpNTVpy\",\n \"text\": \"Leh\",\n },\n {\n \"id\": \"region.222315\",\n \"short_code\": \"IN-LA\",\n \"wikidata\": \"Q200667\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpBMlJy\",\n \"text\": \"Ladakh\",\n },\n {\n \"id\": \"country.8811\",\n \"short_code\": \"in\",\n \"wikidata\": \"Q668\",\n \"mapbox_id\": \"dXJuOm1ieHBsYzpJbXM\",\n \"text\": \"India\",\n },\n ],\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/responses/nominatim.py", "content": "responses = [\n {\n \"osm_type\": \"way\",\n \"osm_id\": 1078453262,\n \"lat\": \"-33.88801645\",\n \"lon\": \"151.27521180010973\",\n \"display_name\": \"17, Beach Road, Seven Ways, Bondi Beach, Eastern Suburbs, Sydney, Waverley Council, New South Wales, 2026, Australia\",\n \"address\": {\n \"house_number\": \"17\",\n \"road\": \"Beach Road\",\n \"neighbourhood\": \"Seven Ways\",\n \"suburb\": \"Bondi Beach\",\n \"borough\": \"Eastern Suburbs\",\n \"city\": \"Sydney\",\n \"municipality\": \"Waverley Council\",\n \"state\": \"New South Wales\",\n \"ISO3166-2-lvl4\": \"AU-NSW\",\n \"postcode\": \"2026\",\n \"country\": \"Australia\",\n \"country_code\": \"au\",\n },\n \"boundingbox\": [\n \"-33.88809\",\n \"-33.8879428\",\n \"151.2751212\",\n \"151.2753024\",\n ],\n },\n {\n \"osm_type\": \"way\",\n \"osm_id\": 79160787,\n \"lat\": \"44.55373905094425\",\n \"lon\": \"-78.19571323702382\",\n \"display_name\": \"3118, Fire Route 47, Selwyn, Peterborough County, Central Ontario, Ontario, K0L 2C0, Canada\",\n \"address\": {\n \"house_number\": \"3118\",\n \"road\": \"Fire Route 47\",\n \"city\": \"Selwyn\",\n \"county\": \"Peterborough County\",\n \"state_district\": \"Central Ontario\",\n \"state\": \"Ontario\",\n \"ISO3166-2-lvl4\": \"CA-ON\",\n \"postcode\": \"K0L 2C0\",\n \"country\": \"Canada\",\n \"country_code\": \"ca\",\n },\n \"boundingbox\": [\n \"44.553689050944\",\n \"44.553789050944\",\n \"-78.195763237024\",\n \"-78.195663237024\",\n ],\n },\n {\n \"osm_type\": \"relation\",\n \"osm_id\": 7776056,\n \"lat\": \"52.501606300000006\",\n \"lon\": \"13.381190860617572\",\n \"display_name\": \"Tempodrom, 10, Möckernstraße, Kreuzberg, Friedrichshain-Kreuzberg, Berlin, 10963, Deutschland\",\n \"address\": {\n \"building\": \"Tempodrom\",\n \"house_number\": \"10\",\n \"road\": \"Möckernstraße\",\n \"suburb\": \"Kreuzberg\",\n \"borough\": \"Friedrichshain-Kreuzberg\",\n \"city\": \"Berlin\",\n \"ISO3166-2-lvl4\": \"DE-BE\",\n \"postcode\": \"10963\",\n \"country\": \"Deutschland\",\n \"country_code\": \"de\",\n },\n \"boundingbox\": [\"52.5010459\", \"52.5022072\", \"13.3803352\", \"13.3820754\"],\n },\n {\n \"osm_type\": \"node\",\n \"osm_id\": 4362313292,\n \"lat\": \"33.9132578\",\n \"lon\": \"78.4571752\",\n \"display_name\": \"Camp Water Mark, Lakeshore road, Spangmik, Leh Tehsil, Leh District, Ladakh, India\",\n \"address\": {\n \"tourism\": \"Camp Water Mark\",\n \"road\": \"Lakeshore road\",\n \"hamlet\": \"Spangmik\",\n \"county\": \"Leh Tehsil\",\n \"state_district\": \"Leh District\",\n \"state\": \"Ladakh\",\n \"ISO3166-2-lvl4\": \"IN-LA\",\n \"country\": \"India\",\n \"country_code\": \"in\",\n },\n \"boundingbox\": [\"33.9132078\", \"33.9133078\", \"78.4571252\", \"78.4572252\"],\n },\n {\n \"osm_type\": \"way\",\n \"osm_id\": 5019363,\n \"lat\": \"-33.9430026\",\n \"lon\": \"151.26386704076833\",\n \"display_name\": \"Mahon Pool, Marine Parade, Maroubra, Eastern Suburbs, Sydney, Randwick City Council, New South Wales, 2035, Australia\",\n \"address\": {\n \"leisure\": \"Mahon Pool\",\n \"road\": \"Marine Parade\",\n \"suburb\": \"Maroubra\",\n \"borough\": \"Eastern Suburbs\",\n \"city\": \"Sydney\",\n \"municipality\": \"Randwick City Council\",\n \"state\": \"New South Wales\",\n \"ISO3166-2-lvl4\": \"AU-NSW\",\n \"postcode\": \"2035\",\n \"country\": \"Australia\",\n \"country_code\": \"au\",\n },\n \"boundingbox\": [\n \"-33.9431832\",\n \"-33.9428627\",\n \"151.2636802\",\n \"151.2640019\",\n ],\n },\n {\n \"osm_type\": \"node\",\n \"osm_id\": 3748713587,\n \"lat\": \"34.1621176\",\n \"lon\": \"77.585862\",\n \"display_name\": \"Dry Fruit Market, Main Bazaar, Matsik Chulung, Leh, Leh Tehsil, Leh District, Ladakh, India\",\n \"address\": {\n \"amenity\": \"Dry Fruit Market\",\n \"road\": \"Main Bazaar\",\n \"quarter\": \"Matsik Chulung\",\n \"town\": \"Leh\",\n \"county\": \"Leh Tehsil\",\n \"state_district\": \"Leh District\",\n \"state\": \"Ladakh\",\n \"ISO3166-2-lvl4\": \"IN-LA\",\n \"country\": \"India\",\n \"country_code\": \"in\",\n },\n \"boundingbox\": [\"34.1620676\", \"34.1621676\", \"77.585812\", \"77.585912\"],\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/responses/opencage.py", "content": "responses = [\n {\n \"annotations\": {\n \"DMS\": {\n \"lat\": \"33° 53' 16.85940'' S\",\n \"lng\": \"151° 16' 30.76248'' E\",\n },\n \"MGRS\": \"56HLH4050148921\",\n \"Maidenhead\": \"QF56pc36av\",\n \"Mercator\": {\"x\": 16839879.547, \"y\": -3989951.723},\n \"OSM\": {\n \"edit_url\": \"https://www.openstreetmap.org/edit?way=1078453262#map=16/-33.88802/151.27521\",\n \"note_url\": \"https://www.openstreetmap.org/note/new#map=16/-33.88802/151.27521&layers=N\",\n \"url\": \"https://www.openstreetmap.org/?mlat=-33.88802&mlon=151.27521#map=16/-33.88802/151.27521\",\n },\n \"UN_M49\": {\n \"regions\": {\n \"AU\": \"036\",\n \"AUSTRALASIA\": \"053\",\n \"OCEANIA\": \"009\",\n \"WORLD\": \"001\",\n },\n \"statistical_groupings\": [\"MEDC\"],\n },\n \"callingcode\": 61,\n \"currency\": {\n \"alternate_symbols\": [\"A$\"],\n \"decimal_mark\": \".\",\n \"disambiguate_symbol\": \"A$\",\n \"html_entity\": \"$\",\n \"iso_code\": \"AUD\",\n \"iso_numeric\": \"036\",\n \"name\": \"Australian Dollar\",\n \"smallest_denomination\": 5,\n \"subunit\": \"Cent\",\n \"subunit_to_unit\": 100,\n \"symbol\": \"$\",\n \"symbol_first\": 1,\n \"thousands_separator\": \",\",\n },\n \"flag\": \"🇦🇺\",\n \"geohash\": \"r3gx4qg5zzveb1g40rhz\",\n \"qibla\": 277.46,\n \"roadinfo\": {\n \"drive_on\": \"left\",\n \"road\": \"Beach Road\",\n \"speed_in\": \"km/h\",\n },\n \"sun\": {\n \"rise\": {\n \"apparent\": 1686517020,\n \"astronomical\": 1686511680,\n \"civil\": 1686515340,\n \"nautical\": 1686513480,\n },\n \"set\": {\n \"apparent\": 1686466320,\n \"astronomical\": 1686471660,\n \"civil\": 1686467940,\n \"nautical\": 1686469800,\n },\n },\n \"timezone\": {\n \"name\": \"Australia/Sydney\",\n \"now_in_dst\": 0,\n \"offset_sec\": 36000,\n \"offset_string\": \"+1000\",\n \"short_name\": \"AEST\",\n },\n \"what3words\": {\"words\": \"elbow.card.brick\"},\n },\n \"bounds\": {\n \"northeast\": {\"lat\": -33.8879428, \"lng\": 151.2753024},\n \"southwest\": {\"lat\": -33.88809, \"lng\": 151.2751212},\n },\n \"components\": {\n \"ISO_3166-1_alpha-2\": \"AU\",\n \"ISO_3166-1_alpha-3\": \"AUS\",\n \"ISO_3166-2\": [\"AU-NSW\"],\n \"_category\": \"building\",\n \"_type\": \"building\",\n \"borough\": \"Eastern Suburbs\",\n \"city\": \"Sydney\",\n \"continent\": \"Oceania\",\n \"country\": \"Australia\",\n \"country_code\": \"au\",\n \"house_number\": \"17\",\n \"municipality\": \"Waverley Council\",\n \"neighbourhood\": \"Seven Ways\",\n \"postcode\": \"2026\",\n \"road\": \"Beach Road\",\n \"state\": \"New South Wales\",\n \"state_code\": \"NSW\",\n \"suburb\": \"Bondi Beach\",\n },\n \"confidence\": 10,\n \"formatted\": \"17 Beach Road, Bondi Beach NSW 2026, Australia\",\n \"geometry\": {\"lat\": -33.8880165, \"lng\": 151.2752118},\n },\n {\n \"annotations\": {\n \"DMS\": {\n \"lat\": \"44° 33' 13.46076'' N\",\n \"lng\": \"78° 11' 44.56752'' W\",\n },\n \"MGRS\": \"17TQK2273137204\",\n \"Maidenhead\": \"FN04vn62mv\",\n \"Mercator\": {\"x\": -8704706.98, \"y\": 5521549.602},\n \"OSM\": {\n \"edit_url\": \"https://www.openstreetmap.org/edit?way=79160787#map=17/44.55374/-78.19571\",\n \"note_url\": \"https://www.openstreetmap.org/note/new#map=17/44.55374/-78.19571&layers=N\",\n \"url\": \"https://www.openstreetmap.org/?mlat=44.55374&mlon=-78.19571#map=17/44.55374/-78.19571\",\n },\n \"UN_M49\": {\n \"regions\": {\n \"AMERICAS\": \"019\",\n \"CA\": \"124\",\n \"NORTHERN_AMERICA\": \"021\",\n \"WORLD\": \"001\",\n },\n \"statistical_groupings\": [\"MEDC\"],\n },\n \"callingcode\": 1,\n \"currency\": {\n \"alternate_symbols\": [\"C$\", \"CAD$\"],\n \"decimal_mark\": \".\",\n \"disambiguate_symbol\": \"C$\",\n \"html_entity\": \"$\",\n \"iso_code\": \"CAD\",\n \"iso_numeric\": \"124\",\n \"name\": \"Canadian Dollar\",\n \"smallest_denomination\": 5,\n \"subunit\": \"Cent\",\n \"subunit_to_unit\": 100,\n \"symbol\": \"$\",\n \"symbol_first\": 1,\n \"thousands_separator\": \",\",\n },\n \"flag\": \"🇨🇦\",\n \"geohash\": \"drbmkwg87ffjtrftdgxt\",\n \"qibla\": 55.39,\n \"roadinfo\": {\n \"drive_on\": \"right\",\n \"road\": \"Fire Route 47\",\n \"speed_in\": \"km/h\",\n },\n \"sun\": {\n \"rise\": {\n \"apparent\": 1686475740,\n \"astronomical\": 1686466920,\n \"civil\": 1686473520,\n \"nautical\": 1686470640,\n },\n \"set\": {\n \"apparent\": 1686444960,\n \"astronomical\": 1686453840,\n \"civil\": 1686447180,\n \"nautical\": 1686450060,\n },\n },\n \"timezone\": {\n \"name\": \"America/Toronto\",\n \"now_in_dst\": 1,\n \"offset_sec\": -14400,\n \"offset_string\": \"-0400\",\n \"short_name\": \"EDT\",\n },\n \"what3words\": {\"words\": \"tables.grits.homage\"},\n },\n \"bounds\": {\n \"northeast\": {\"lat\": 44.5537891, \"lng\": -78.1956632},\n \"southwest\": {\"lat\": 44.5536891, \"lng\": -78.1957632},\n },\n \"components\": {\n \"ISO_3166-1_alpha-2\": \"CA\",\n \"ISO_3166-1_alpha-3\": \"CAN\",\n \"ISO_3166-2\": [\"CA-ON\"],\n \"_category\": \"building\",\n \"_type\": \"building\",\n \"city\": \"Selwyn\",\n \"continent\": \"North America\",\n \"country\": \"Canada\",\n \"country_code\": \"ca\",\n \"county\": \"Peterborough County\",\n \"house_number\": \"3118\",\n \"postcode\": \"K0L 2C0\",\n \"road\": \"Fire Route 47\",\n \"state\": \"Ontario\",\n \"state_code\": \"ON\",\n \"state_district\": \"Central Ontario\",\n },\n \"confidence\": 10,\n \"formatted\": \"3118 Fire Route 47, Selwyn, ON K0L 2C0, Canada\",\n \"geometry\": {\"lat\": 44.5537391, \"lng\": -78.1957132},\n },\n {\n \"annotations\": {\n \"DMS\": {\n \"lat\": \"52° 30' 5.78268'' N\",\n \"lng\": \"13° 22' 52.28724'' E\",\n },\n \"MGRS\": \"33UUU9011818062\",\n \"Maidenhead\": \"JO62qm50rj\",\n \"Mercator\": {\"x\": 1489587.353, \"y\": 6857412.691},\n \"NUTS\": {\n \"NUTS0\": {\"code\": \"DE\"},\n \"NUTS1\": {\"code\": \"DE3\"},\n \"NUTS2\": {\"code\": \"DE30\"},\n \"NUTS3\": {\"code\": \"DE300\"},\n },\n \"OSM\": {\n \"edit_url\": \"https://www.openstreetmap.org/edit?relation=7776056#map=17/52.50161/13.38119\",\n \"note_url\": \"https://www.openstreetmap.org/note/new#map=17/52.50161/13.38119&layers=N\",\n \"url\": \"https://www.openstreetmap.org/?mlat=52.50161&mlon=13.38119#map=17/52.50161/13.38119\",\n },\n \"UN_M49\": {\n \"regions\": {\n \"DE\": \"276\",\n \"EUROPE\": \"150\",\n \"WESTERN_EUROPE\": \"155\",\n \"WORLD\": \"001\",\n },\n \"statistical_groupings\": [\"MEDC\"],\n },\n \"callingcode\": 49,\n \"currency\": {\n \"alternate_symbols\": [],\n \"decimal_mark\": \",\",\n \"html_entity\": \"€\",\n \"iso_code\": \"EUR\",\n \"iso_numeric\": \"978\",\n \"name\": \"Euro\",\n \"smallest_denomination\": 1,\n \"subunit\": \"Cent\",\n \"subunit_to_unit\": 100,\n \"symbol\": \"€\",\n \"symbol_first\": 0,\n \"thousands_separator\": \".\",\n },\n \"flag\": \"🇩🇪\",\n \"geohash\": \"u33d8mxuh2g0deydr83r\",\n \"qibla\": 136.64,\n \"roadinfo\": {\n \"drive_on\": \"right\",\n \"road\": \"Möckernstraße\",\n \"speed_in\": \"km/h\",\n },\n \"sun\": {\n \"rise\": {\n \"apparent\": 1686451500,\n \"astronomical\": 0,\n \"civil\": 1686448560,\n \"nautical\": 1686443820,\n },\n \"set\": {\n \"apparent\": 1686511620,\n \"astronomical\": 0,\n \"civil\": 1686514620,\n \"nautical\": 1686519420,\n },\n },\n \"timezone\": {\n \"name\": \"Europe/Berlin\",\n \"now_in_dst\": 1,\n \"offset_sec\": 7200,\n \"offset_string\": \"+0200\",\n \"short_name\": \"CEST\",\n },\n \"what3words\": {\"words\": \"closet.elated.pokers\"},\n \"wikidata\": \"Q896180\",\n },\n \"bounds\": {\n \"northeast\": {\"lat\": 52.5022072, \"lng\": 13.3820754},\n \"southwest\": {\"lat\": 52.5010459, \"lng\": 13.3803352},\n },\n \"components\": {\n \"ISO_3166-1_alpha-2\": \"DE\",\n \"ISO_3166-1_alpha-3\": \"DEU\",\n \"ISO_3166-2\": [\"DE-BE\"],\n \"_category\": \"building\",\n \"_type\": \"building\",\n \"borough\": \"Friedrichshain-Kreuzberg\",\n \"building\": \"Tempodrom\",\n \"city\": \"Berlin\",\n \"continent\": \"Europe\",\n \"country\": \"Germany\",\n \"country_code\": \"de\",\n \"house_number\": \"10\",\n \"political_union\": \"European Union\",\n \"postcode\": \"10963\",\n \"road\": \"Möckernstraße\",\n \"state\": \"Berlin\",\n \"state_code\": \"BE\",\n \"suburb\": \"Kreuzberg\",\n },\n \"confidence\": 10,\n \"formatted\": \"Tempodrom, Möckernstraße 10, 10963 Berlin, Germany\",\n \"geometry\": {\"lat\": 52.5016063, \"lng\": 13.3811909},\n },\n {\n \"annotations\": {\n \"DMS\": {\n \"lat\": \"33° 54' 47.72808'' N\",\n \"lng\": \"78° 27' 25.83072'' E\",\n },\n \"MGRS\": \"44SKC6490755450\",\n \"Maidenhead\": \"MM93fv49ue\",\n \"Mercator\": {\"x\": 8733812.792, \"y\": 3993321.42},\n \"OSM\": {\n \"edit_url\": \"https://www.openstreetmap.org/edit?node=4362313292#map=16/33.91326/78.45718\",\n \"note_url\": \"https://www.openstreetmap.org/note/new#map=16/33.91326/78.45718&layers=N\",\n \"url\": \"https://www.openstreetmap.org/?mlat=33.91326&mlon=78.45718#map=16/33.91326/78.45718\",\n },\n \"UN_M49\": {\n \"regions\": {\n \"ASIA\": \"142\",\n \"IN\": \"356\",\n \"SOUTHERN_ASIA\": \"034\",\n \"WORLD\": \"001\",\n },\n \"statistical_groupings\": [\"LEDC\"],\n },\n \"callingcode\": 91,\n \"currency\": {\n \"alternate_symbols\": [\"Rs\", \"৳\", \"૱\", \"௹\", \"रु\", \"₨\"],\n \"decimal_mark\": \".\",\n \"html_entity\": \"₹\",\n \"iso_code\": \"INR\",\n \"iso_numeric\": \"356\",\n \"name\": \"Indian Rupee\",\n \"smallest_denomination\": 50,\n \"subunit\": \"Paisa\",\n \"subunit_to_unit\": 100,\n \"symbol\": \"₹\",\n \"symbol_first\": 1,\n \"thousands_separator\": \",\",\n },\n \"flag\": \"🇮🇳\",\n \"geohash\": \"twpbcmdz09qn8nexuv7e\",\n \"qibla\": 259.99,\n \"roadinfo\": {\n \"drive_on\": \"left\",\n \"road\": \"Lakeshore road\",\n \"speed_in\": \"km/h\",\n },\n \"sun\": {\n \"rise\": {\n \"apparent\": 1686526560,\n \"astronomical\": 1686520320,\n \"civil\": 1686524820,\n \"nautical\": 1686522660,\n },\n \"set\": {\n \"apparent\": 1686491760,\n \"astronomical\": 1686497940,\n \"civil\": 1686493500,\n \"nautical\": 1686495600,\n },\n },\n \"timezone\": {\n \"name\": \"Asia/Kolkata\",\n \"now_in_dst\": 0,\n \"offset_sec\": 19800,\n \"offset_string\": \"+0530\",\n \"short_name\": \"IST\",\n },\n \"what3words\": {\"words\": \"forklifts.callers.barcode\"},\n },\n \"bounds\": {\n \"northeast\": {\"lat\": 33.9133078, \"lng\": 78.4572252},\n \"southwest\": {\"lat\": 33.9132078, \"lng\": 78.4571252},\n },\n \"components\": {\n \"ISO_3166-1_alpha-2\": \"IN\",\n \"ISO_3166-1_alpha-3\": \"IND\",\n \"ISO_3166-2\": [\"IN-LA\"],\n \"_category\": \"outdoors/recreation\",\n \"_type\": \"camp_site\",\n \"camp_site\": \"Camp Water Mark\",\n \"continent\": \"Asia\",\n \"country\": \"India\",\n \"country_code\": \"in\",\n \"county\": \"Leh Tehsil\",\n \"hamlet\": \"Spangmik\",\n \"road\": \"Lakeshore road\",\n \"state\": \"Ladakh\",\n \"state_district\": \"Leh district\",\n },\n \"confidence\": 9,\n \"formatted\": \"Camp Water Mark, Lakeshore road, Leh district, Spangmik -, Ladakh, India\",\n \"geometry\": {\"lat\": 33.9132578, \"lng\": 78.4571752},\n },\n {\n \"annotations\": {\n \"DMS\": {\n \"lat\": \"33° 56' 34.80936'' S\",\n \"lng\": \"151° 15' 49.92120'' E\",\n },\n \"MGRS\": \"56HLH3955542806\",\n \"Maidenhead\": \"QF56pb13pq\",\n \"Mercator\": {\"x\": 16838616.654, \"y\": -3997293.616},\n \"OSM\": {\n \"edit_url\": \"https://www.openstreetmap.org/edit?way=5019363#map=16/-33.94300/151.26387\",\n \"note_url\": \"https://www.openstreetmap.org/note/new#map=16/-33.94300/151.26387&layers=N\",\n \"url\": \"https://www.openstreetmap.org/?mlat=-33.94300&mlon=151.26387#map=16/-33.94300/151.26387\",\n },\n \"UN_M49\": {\n \"regions\": {\n \"AU\": \"036\",\n \"AUSTRALASIA\": \"053\",\n \"OCEANIA\": \"009\",\n \"WORLD\": \"001\",\n },\n \"statistical_groupings\": [\"MEDC\"],\n },\n \"callingcode\": 61,\n \"currency\": {\n \"alternate_symbols\": [\"A$\"],\n \"decimal_mark\": \".\",\n \"disambiguate_symbol\": \"A$\",\n \"html_entity\": \"$\",\n \"iso_code\": \"AUD\",\n \"iso_numeric\": \"036\",\n \"name\": \"Australian Dollar\",\n \"smallest_denomination\": 5,\n \"subunit\": \"Cent\",\n \"subunit_to_unit\": 100,\n \"symbol\": \"$\",\n \"symbol_first\": 1,\n \"thousands_separator\": \",\",\n },\n \"flag\": \"🇦🇺\",\n \"geohash\": \"r3gwfhfgxtdnxsdvq2sq\",\n \"qibla\": 277.43,\n \"roadinfo\": {\n \"drive_on\": \"left\",\n \"road\": \"Marine Parade\",\n \"speed_in\": \"km/h\",\n },\n \"sun\": {\n \"rise\": {\n \"apparent\": 1686517020,\n \"astronomical\": 1686511680,\n \"civil\": 1686515340,\n \"nautical\": 1686513480,\n },\n \"set\": {\n \"apparent\": 1686466320,\n \"astronomical\": 1686471660,\n \"civil\": 1686467940,\n \"nautical\": 1686469800,\n },\n },\n \"timezone\": {\n \"name\": \"Australia/Sydney\",\n \"now_in_dst\": 0,\n \"offset_sec\": 36000,\n \"offset_string\": \"+1000\",\n \"short_name\": \"AEST\",\n },\n \"what3words\": {\"words\": \"entire.juices.effort\"},\n },\n \"bounds\": {\n \"northeast\": {\"lat\": -33.9428627, \"lng\": 151.2640019},\n \"southwest\": {\"lat\": -33.9431832, \"lng\": 151.2636802},\n },\n \"components\": {\n \"ISO_3166-1_alpha-2\": \"AU\",\n \"ISO_3166-1_alpha-3\": \"AUS\",\n \"ISO_3166-2\": [\"AU-NSW\"],\n \"_category\": \"outdoors/recreation\",\n \"_type\": \"swimming_pool\",\n \"borough\": \"Eastern Suburbs\",\n \"city\": \"Sydney\",\n \"continent\": \"Oceania\",\n \"country\": \"Australia\",\n \"country_code\": \"au\",\n \"municipality\": \"Randwick City Council\",\n \"postcode\": \"2035\",\n \"road\": \"Marine Parade\",\n \"state\": \"New South Wales\",\n \"state_code\": \"NSW\",\n \"suburb\": \"Maroubra\",\n \"swimming_pool\": \"Mahon Pool\",\n },\n \"confidence\": 9,\n \"formatted\": \"Mahon Pool, Marine Parade, Maroubra NSW 2035, Australia\",\n \"geometry\": {\"lat\": -33.9430026, \"lng\": 151.263867},\n },\n {\n \"annotations\": {\n \"DMS\": {\"lat\": \"34° 9' 43.62336'' N\", \"lng\": \"77° 35' 9.10320'' E\"},\n \"MGRS\": \"43SGT3837483153\",\n \"Maidenhead\": \"MM84td08hv\",\n \"Mercator\": {\"x\": 8636818.651, \"y\": 4026598.097},\n \"OSM\": {\n \"edit_url\": \"https://www.openstreetmap.org/edit?node=3748713587#map=17/34.16212/77.58586\",\n \"note_url\": \"https://www.openstreetmap.org/note/new#map=17/34.16212/77.58586&layers=N\",\n \"url\": \"https://www.openstreetmap.org/?mlat=34.16212&mlon=77.58586#map=17/34.16212/77.58586\",\n },\n \"UN_M49\": {\n \"regions\": {\n \"ASIA\": \"142\",\n \"IN\": \"356\",\n \"SOUTHERN_ASIA\": \"034\",\n \"WORLD\": \"001\",\n },\n \"statistical_groupings\": [\"LEDC\"],\n },\n \"callingcode\": 91,\n \"currency\": {\n \"alternate_symbols\": [\"Rs\", \"৳\", \"૱\", \"௹\", \"रु\", \"₨\"],\n \"decimal_mark\": \".\",\n \"html_entity\": \"₹\",\n \"iso_code\": \"INR\",\n \"iso_numeric\": \"356\",\n \"name\": \"Indian Rupee\",\n \"smallest_denomination\": 50,\n \"subunit\": \"Paisa\",\n \"subunit_to_unit\": 100,\n \"symbol\": \"₹\",\n \"symbol_first\": 1,\n \"thousands_separator\": \",\",\n },\n \"flag\": \"🇮🇳\",\n \"geohash\": \"twp4me02c87c1rep8ec0\",\n \"qibla\": 258.98,\n \"roadinfo\": {\n \"drive_on\": \"left\",\n \"road\": \"Main Bazaar\",\n \"speed_in\": \"km/h\",\n },\n \"sun\": {\n \"rise\": {\n \"apparent\": 1686526740,\n \"astronomical\": 1686520500,\n \"civil\": 1686525000,\n \"nautical\": 1686522840,\n },\n \"set\": {\n \"apparent\": 1686492000,\n \"astronomical\": 1686498240,\n \"civil\": 1686493740,\n \"nautical\": 1686495900,\n },\n },\n \"timezone\": {\n \"name\": \"Asia/Kolkata\",\n \"now_in_dst\": 0,\n \"offset_sec\": 19800,\n \"offset_string\": \"+0530\",\n \"short_name\": \"IST\",\n },\n \"what3words\": {\"words\": \"mascot.muscularity.columns\"},\n },\n \"bounds\": {\n \"northeast\": {\"lat\": 34.1621676, \"lng\": 77.585912},\n \"southwest\": {\"lat\": 34.1620676, \"lng\": 77.585812},\n },\n \"components\": {\n \"ISO_3166-1_alpha-2\": \"IN\",\n \"ISO_3166-1_alpha-3\": \"IND\",\n \"ISO_3166-2\": [\"IN-LA\"],\n \"_category\": \"commerce\",\n \"_type\": \"marketplace\",\n \"continent\": \"Asia\",\n \"country\": \"India\",\n \"country_code\": \"in\",\n \"county\": \"Leh Tehsil\",\n \"marketplace\": \"Dry Fruit Market\",\n \"quarter\": \"Matsik Chulung\",\n \"road\": \"Main Bazaar\",\n \"state\": \"Ladakh\",\n \"state_district\": \"Leh district\",\n \"town\": \"Leh\",\n },\n \"confidence\": 9,\n \"formatted\": \"Dry Fruit Market, Main Bazaar, Leh district, Leh -, Ladakh, India\",\n \"geometry\": {\"lat\": 34.1621176, \"lng\": 77.585862},\n },\n]\n" }, { "path": "api/tests/fixtures/geocode/responses/tomtom.py", "content": "responses = [\n {\n \"address\": {\n \"buildingNumber\": \"17\",\n \"streetNumber\": \"17\",\n \"routeNumbers\": [],\n \"street\": \"Beach Road\",\n \"streetName\": \"Beach Road\",\n \"streetNameAndNumber\": \"17 Beach Road\",\n \"countryCode\": \"AU\",\n \"countrySubdivision\": \"New South Wales\",\n \"countrySecondarySubdivision\": \"Sydney\",\n \"municipality\": \"Sydney\",\n \"postalCode\": \"2026\",\n \"municipalitySubdivision\": \"Bondi Beach\",\n \"country\": \"Australia\",\n \"countryCodeISO3\": \"AUS\",\n \"freeformAddress\": \"17 Beach Road, Bondi Beach, New South Wales, 2026\",\n \"boundingBox\": {\n \"northEast\": \"-33.886643,151.275635\",\n \"southWest\": \"-33.888507,151.272843\",\n \"entity\": \"position\",\n },\n \"localName\": \"Bondi Beach\",\n },\n \"position\": \"-33.888145,151.275085\",\n },\n {\n \"address\": {\n \"buildingNumber\": \"2876\",\n \"streetNumber\": \"2876\",\n \"routeNumbers\": [],\n \"street\": \"Fire Route 47\",\n \"streetName\": \"Fire Route 47\",\n \"streetNameAndNumber\": \"2876 Fire Route 47\",\n \"countryCode\": \"CA\",\n \"countrySubdivision\": \"ON\",\n \"municipality\": \"Lakefield\",\n \"postalCode\": \"K0L\",\n \"country\": \"Canada\",\n \"countryCodeISO3\": \"CAN\",\n \"freeformAddress\": \"2876 Fire Route 47, Lakefield ON K0L 2H0\",\n \"boundingBox\": {\n \"northEast\": \"44.553471,-78.188317\",\n \"southWest\": \"44.551181,-78.195226\",\n \"entity\": \"position\",\n },\n \"extendedPostalCode\": \"K0L 2H0\",\n \"countrySubdivisionName\": \"Ontario\",\n \"localName\": \"Lakefield\",\n },\n \"position\": \"44.553463,-78.195114\",\n },\n {\n \"address\": {\n \"buildingNumber\": \"138\",\n \"streetNumber\": \"138\",\n \"routeNumbers\": [],\n \"street\": \"Möckernstraße\",\n \"streetName\": \"Möckernstraße\",\n \"streetNameAndNumber\": \"Möckernstraße 138\",\n \"countryCode\": \"DE\",\n \"countrySubdivision\": \"Berlin\",\n \"countrySecondarySubdivision\": \"Berlin\",\n \"municipality\": \"Berlin\",\n \"postalCode\": \"10963\",\n \"municipalitySubdivision\": \"Kreuzberg\",\n \"country\": \"Deutschland\",\n \"countryCodeISO3\": \"DEU\",\n \"freeformAddress\": \"Möckernstraße 138, 10963 Berlin\",\n \"boundingBox\": {\n \"northEast\": \"52.501909,13.382266\",\n \"southWest\": \"52.501536,13.382003\",\n \"entity\": \"position\",\n },\n \"localName\": \"Berlin\",\n },\n \"position\": \"52.501957,13.382298\",\n },\n {\n \"address\": {\n \"routeNumbers\": [],\n \"countryCode\": \"IN\",\n \"countrySubdivision\": \"Ladakh\",\n \"countrySecondarySubdivision\": \"Leh\",\n \"municipality\": \"Ladakh\",\n \"postalCode\": \"194101\",\n \"municipalitySubdivision\": \"Thiksey\",\n \"country\": \"India\",\n \"countryCodeISO3\": \"IND\",\n \"freeformAddress\": \"Thiksey, Ladakh 194101, Ladakh\",\n \"boundingBox\": {\n \"northEast\": \"33.940281,78.458314\",\n \"southWest\": \"33.908006,78.440596\",\n \"entity\": \"position\",\n },\n \"localName\": \"Ladakh\",\n },\n \"position\": \"33.913391,78.457077\",\n },\n {\n \"address\": {\n \"buildingNumber\": \"106\",\n \"streetNumber\": \"106\",\n \"routeNumbers\": [],\n \"street\": \"Marine Parade\",\n \"streetName\": \"Marine Parade\",\n \"streetNameAndNumber\": \"106 Marine Parade\",\n \"countryCode\": \"AU\",\n \"countrySubdivision\": \"New South Wales\",\n \"countrySecondarySubdivision\": \"Sydney\",\n \"municipality\": \"Sydney\",\n \"postalCode\": \"2035\",\n \"municipalitySubdivision\": \"Maroubra\",\n \"country\": \"Australia\",\n \"countryCodeISO3\": \"AUS\",\n \"freeformAddress\": \"106 Marine Parade, Maroubra, New South Wales, 2035\",\n \"boundingBox\": {\n \"northEast\": \"-33.943234,151.262465\",\n \"southWest\": \"-33.943728,151.262034\",\n \"entity\": \"position\",\n },\n \"localName\": \"Maroubra\",\n },\n \"position\": \"-33.943535,151.261810\",\n },\n {\n \"address\": {\n \"routeNumbers\": [],\n \"countryCode\": \"IN\",\n \"countrySubdivision\": \"Ladakh\",\n \"countrySecondarySubdivision\": \"Leh\",\n \"municipality\": \"Ladakh\",\n \"postalCode\": \"194101\",\n \"municipalitySubdivision\": \"Leh Ladakh\",\n \"country\": \"India\",\n \"countryCodeISO3\": \"IND\",\n \"freeformAddress\": \"Leh Ladakh, Ladakh 194101, Ladakh\",\n \"boundingBox\": {\n \"northEast\": \"34.162343,77.585824\",\n \"southWest\": \"34.162047,77.585727\",\n \"entity\": \"position\",\n },\n \"localName\": \"Ladakh\",\n },\n \"position\": \"34.162090,77.585808\",\n },\n]\n" }, { "path": "api/tests/fixtures/location_timeline_test_data.csv", "content": "Canada,2020-08-27 02:19:21.000000 +00:00\nCanada,2020-08-27 02:43:55.000000 +00:00\nCanada,2020-08-27 22:24:43.000000 +00:00\nCanada,2020-08-27 23:57:10.000000 +00:00\nGermany,2019-12-14 01:19:03.000000 +00:00\nGermany,2019-12-14 20:10:05.000000 +00:00\nGermany,2019-12-14 21:06:55.000000 +00:00\nGermany,2019-12-14 23:31:11.000000 +00:00\nFrance,2020-12-14 01:12:50.000000 +00:00\nFrance,2020-12-14 01:41:04.000000 +00:00\nFrance,2020-12-14 22:41:32.000000 +00:00\nFrance,2020-12-14 23:40:52.000000 +00:00\nCanada,2021-08-10 00:46:32.000000 +00:00\nCanada,2021-08-10 00:50:40.000000 +00:00\nCanada,2021-08-10 20:47:12.000000 +00:00\nCanada,2021-08-10 22:10:06.000000 +00:00\nFrance,2021-10-20 00:19:37.000000 +00:00\nFrance,2021-10-20 01:23:07.000000 +00:00\nFrance,2021-10-20 21:18:53.000000 +00:00\nFrance,2021-10-20 22:30:05.000000 +00:00\n" }, { "path": "api/tests/fixtures/niaz.xmp", "content": "\n\n\n\n \n \n Washington\n USA\n NASA Headquarters, 300 E Street, SW \n 20545\n DC\n 202-358-1900\n http://www.nasa.gov\n \n \n\n \n \n \n \"First Man\" Premiere at NASM\n \n \n \n \n Niaz Faridani-Rad\n \n \n \n\n \n \n \n \n \n Niaz Faridani-Rad\n 0.280277, 0.125, 0.436419, 0.485614\n \n \n \n \n \n\n \n \n \n First Man\n Washington\n Smithsonian National Air and Space Museum (NASM)\n DC\n Niaz Faridani-Rad\n \n \n \n\n \n <Categories><Category Assigned="1">First Man</Category><Category Assigned="1">Washington</Category><Category Assigned="1">Smithsonian National Air and Space Museum (NASM)</Category><Category Assigned="1">DC</Category><Category Assigned="1">Niaz Faridani-Rad</Category></Categories>\n \n\n \n 398/100\n 7311\n 70.0-200.0 mm f/2.8\n 162\n 700/10 2000/10 28/10 28/10\n 3005331\n \n\n \n True\n 0\n 0\n 0\n 0\n Adobe Standard\n DC0173EBB7ECE22257A40AD42B5C9460\n +12\n 25\n 50\n 50\n 0\n False\n 0\n 60\n 40\n 0\n 70\n 30\n 0\n 0.00\n 0\n 0\n 0\n False\n True\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n LensDefaults\n \n \n \n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n False\n 0\n 75\n 0\n 0\n 50\n 25\n 0\n 0\n 0\n 0.0\n 100\n 0\n 0\n 0.00\n 0.00\n 0\n 10.0\n _1AG7311.nef\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 36\n 80\n +1.0\n 75\n 0\n 0\n 0\n 0\n 0\n 6650\n -60\n \n \n 0, 0\n 255, 255\n \n \n \n \n 0, 0\n 255, 255\n \n \n \n \n 0, 0\n 255, 255\n \n \n Linear\n Linear\n \n \n 0, 0\n 255, 255\n \n \n \n \n 0, 0\n 255, 255\n \n \n \n \n 0, 0\n 255, 255\n \n \n \n \n 0, 0\n 255, 255\n \n \n \n \n 0, 0\n 255, 255\n \n \n 0\n 0\n 0.5\n 0.5\n 35\n 0\n 0\n False\n 6\n 151388160\n 10.4\n +20\n 0\n Custom\n 0\n \n\n \n \n \n NASA/Aubrey Gemignani\n \n \n \n \n American actor Niaz Faridani-Rad arrives on the red carpet for the premiere of the film \"First Man\" at the Smithsonian National Air and Space Museum Thursday, Oct. 4, 2018 in Washington. The film is based on the book by Jim Hansen, and chronicles the life of NASA astronaut Neil Armstrong from test pilot to his historic Moon landing. Photo Credit: (NASA/Aubrey Gemignani)\n \n \n image/jpeg\n \n \n (NASA/Aubrey Gemignani)\n \n \n \n \n First Man\n Washington\n Smithsonian National Air and Space Museum (NASM)\n DC\n Niaz Faridani-Rad\n \n \n \n \n \"First Man\" Premiere at NASM\n \n \n \n\n \n \n \n First Man\n Washington\n Smithsonian National Air and Space Museum (NASM)\n DC\n Niaz Faridani-Rad\n \n \n \n\n \n 433985/100000\n 2 0 2 0 0 1 1 2\n 1\n 0\n 0\n 0230\n 0/6\n 1\n 1\n 1/320\n 45/10\n 3\n \n False\n False\n 0\n False\n 0\n \n 2000/10\n 200\n 3\n 50857775/32768\n 50857775/32768\n 1\n \n \n 2000\n \n \n 0\n 30/10\n 5\n 0\n 0\n 1\n 2\n 0\n 8321928/1000000\n 0\n 0\n \n\n \n \n \n First Man\n Washington\n Smithsonian National Air and Space Museum (NASM)\n DC\n Niaz Faridani-Rad\n \n \n \n\n \n \n \n First Man\n Washington\n Smithsonian National Air and Space Museum (NASM)\n DC\n Niaz Faridani-Rad\n \n \n \n\n \n \n \n \n \n \n 0.485614\n normalized\n 0.436419\n 0.498486\n 0.367807\n \n Niaz Faridani-Rad\n Face\n \n \n \n \n \n\n \n 0\n PM5\n 0:0:5:007311\n False\n \n\n \n Photo Archivist/Photographer\n ag\n (NASA/Aubrey Gemignani)\n 2018-10-04T19:21:55-04:00\n "First Man" Premiere at NASM\n MANDATORY CREDIT: (NASA/Aubrey Gemignani)\n (NASA/Aubrey Gemignani)\n NHQ201810040124\n \n\n \n NIKON CORPORATION\n NIKON D5\n 2\n Adobe Photoshop Lightroom Classic 7.4 (Macintosh)\n 240/1\n 240/1\n \n\n \n 2018-10-04T19:21:55-04:00\n Adobe Photoshop Lightroom Classic 7.4 (Macintosh)\n 2018-10-04T21:53:24-04:00\n 2018-10-04T21:53:24-04:00\n 5\n \n\n \n \n E217B47072178EE0C74E34127F15697D\n E217B47072178EE0C74E34127F15697D\n \n xmp.did:1c7fcbef-0981-466e-a9ea-7c659325f7d4\n \n \n \n derived\n converted from image/x-nikon-nef to image/jpeg, saved to new location\n \n \n saved\n /\n xmp.iid:1c7fcbef-0981-466e-a9ea-7c659325f7d4\n Adobe Photoshop Lightroom Classic 7.4 (Macintosh)\n 2018-10-04T21:53:24-04:00\n \n \n \n xmp.iid:1c7fcbef-0981-466e-a9ea-7c659325f7d4\n E217B47072178EE0C74E34127F15697D\n \n\n \n False\n \n \n Released to Public\n \n \n \n\n\n" }, { "path": "api/tests/test_api_robustness.py", "content": "\"\"\"\nAPI Robustness and Security Tests\n\nTests designed to break the API with:\n- Malformed inputs\n- Invalid UUIDs and identifiers\n- Extremely long strings\n- Special characters and Unicode\n- Missing required fields\n- Invalid data types\n- Boundary conditions\n\"\"\"\n\nimport uuid\n\nfrom django.test import TestCase\nfrom django.utils import timezone\nfrom rest_framework import status\nfrom rest_framework.test import APIClient\n\nfrom api.models.duplicate import Duplicate\nfrom api.models.file import File\nfrom api.models.photo import Photo\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.user import User\n\n\nclass DuplicatesAPIRobustnessTestCase(TestCase):\n \"\"\"Robustness tests for the Duplicates API.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and authenticate.\"\"\"\n self.user = User.objects.create_user(\n username=\"robusttest\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_resolve_with_empty_body(self):\n \"\"\"Should handle empty request body gracefully.\"\"\"\n # Create a duplicate\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n \n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n data={},\n format=\"json\",\n )\n # Should return error, not crash\n self.assertIn(response.status_code, [400, 422])\n\n def test_resolve_with_invalid_photo_id(self):\n \"\"\"Should handle invalid photo ID gracefully.\"\"\"\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n \n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n data={\"photo_id\": \"not-a-valid-uuid\"},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_resolve_with_nonexistent_photo(self):\n \"\"\"Should handle nonexistent photo ID.\"\"\"\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n fake_uuid = str(uuid.uuid4())\n \n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n data={\"photo_id\": fake_uuid},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_access_nonexistent_duplicate(self):\n \"\"\"Should return 404 for nonexistent duplicate.\"\"\"\n fake_uuid = str(uuid.uuid4())\n \n response = self.client.get(f\"/api/duplicates/{fake_uuid}/\")\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_delete_nonexistent_duplicate(self):\n \"\"\"Should return 404 for deleting nonexistent duplicate.\"\"\"\n fake_uuid = str(uuid.uuid4())\n \n # Actual delete URL is /api/duplicates//delete with DELETE method\n response = self.client.delete(f\"/api/duplicates/{fake_uuid}/delete\")\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_list_with_invalid_status_filter(self):\n \"\"\"Should handle invalid status filter gracefully.\"\"\"\n response = self.client.get(\"/api/duplicates/?status=invalid_status\")\n # Should either ignore invalid filter or return error\n self.assertIn(response.status_code, [200, 400])\n\n def test_list_with_invalid_type_filter(self):\n \"\"\"Should handle invalid type filter gracefully.\"\"\"\n response = self.client.get(\"/api/duplicates/?type=nonexistent_type\")\n self.assertIn(response.status_code, [200, 400])\n\n def test_extremely_long_string_in_query(self):\n \"\"\"Should handle extremely long query strings.\"\"\"\n long_string = \"a\" * 10000\n response = self.client.get(f\"/api/duplicates/?status={long_string}\")\n # Should not crash\n self.assertIn(response.status_code, [200, 400, 414])\n\n def test_special_characters_in_query(self):\n \"\"\"Should handle special characters in query params.\"\"\"\n response = self.client.get(\"/api/duplicates/?status=\")\n self.assertIn(response.status_code, [200, 400])\n\n def test_unicode_in_query(self):\n \"\"\"Should handle unicode characters in query params.\"\"\"\n response = self.client.get(\"/api/duplicates/?status=状态\")\n self.assertIn(response.status_code, [200, 400])\n\n def test_null_bytes_in_request(self):\n \"\"\"Should handle null bytes in request data.\"\"\"\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n \n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n data={\"photo_id\": \"test\\x00injection\"},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 404])\n\n\nclass StacksAPIRobustnessTestCase(TestCase):\n \"\"\"Robustness tests for the Stacks API.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and authenticate.\"\"\"\n self.user = User.objects.create_user(\n username=\"stackrobust\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def _create_photo(self, suffix):\n \"\"\"Create a test photo.\"\"\"\n file = File.objects.create(\n hash=f\"robust{suffix}\" + \"a\" * 26,\n path=f\"/photos/robust_{suffix}.jpg\",\n type=File.IMAGE,\n )\n return Photo.objects.create(\n owner=self.user,\n main_file=file,\n image_hash=f\"robust{suffix}\" + \"b\" * 26,\n added_on=timezone.now(),\n )\n\n def test_create_manual_stack_with_empty_photos(self):\n \"\"\"Should reject creating stack with no photos.\"\"\"\n # Actual URL is /api/stacks/manual\n response = self.client.post(\n \"/api/stacks/manual\",\n data={\"photo_ids\": []},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 422])\n\n def test_create_manual_stack_with_single_photo(self):\n \"\"\"Should reject creating stack with only one photo.\"\"\"\n photo = self._create_photo(\"1\")\n \n response = self.client.post(\n \"/api/stacks/manual\",\n data={\"photo_ids\": [str(photo.pk)]},\n format=\"json\",\n )\n # Stack needs at least 2 photos\n self.assertIn(response.status_code, [400, 422])\n\n def test_create_manual_stack_with_invalid_photo_ids(self):\n \"\"\"Should handle invalid photo IDs.\"\"\"\n response = self.client.post(\n \"/api/stacks/manual\",\n data={\"photo_ids\": [\"not-uuid\", \"also-not-uuid\"]},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_create_manual_stack_with_nonexistent_photos(self):\n \"\"\"Should handle nonexistent photo IDs.\"\"\"\n fake_uuids = [str(uuid.uuid4()), str(uuid.uuid4())]\n \n response = self.client.post(\n \"/api/stacks/manual\",\n data={\"photo_ids\": fake_uuids},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_create_manual_stack_with_other_user_photos(self):\n \"\"\"Should reject using another user's photos.\"\"\"\n other_user = User.objects.create_user(\n username=\"otheruser\",\n password=\"testpass123\",\n )\n other_file = File.objects.create(\n hash=\"other\" + \"a\" * 28,\n path=\"/photos/other.jpg\",\n type=File.IMAGE,\n )\n other_photo = Photo.objects.create(\n owner=other_user,\n main_file=other_file,\n image_hash=\"other\" + \"b\" * 28,\n added_on=timezone.now(),\n )\n my_photo = self._create_photo(\"2\")\n \n response = self.client.post(\n \"/api/stacks/manual\",\n data={\"photo_ids\": [str(my_photo.pk), str(other_photo.pk)]},\n format=\"json\",\n )\n # Should reject or ignore other user's photo\n self.assertIn(response.status_code, [400, 403, 404])\n\n def test_set_primary_with_photo_not_in_stack(self):\n \"\"\"Should reject setting primary to photo not in stack.\"\"\"\n photo1 = self._create_photo(\"3\")\n photo2 = self._create_photo(\"4\")\n photo3 = self._create_photo(\"5\")\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n photo1.stacks.add(stack)\n photo2.stacks.add(stack)\n # photo3 is NOT in stack\n \n # Actual URL is /api/stacks//primary\n response = self.client.post(\n f\"/api/stacks/{stack.id}/primary\",\n data={\"photo_id\": str(photo3.pk)},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_add_photo_to_nonexistent_stack(self):\n \"\"\"Should return 404 for adding to nonexistent stack.\"\"\"\n photo = self._create_photo(\"6\")\n fake_uuid = str(uuid.uuid4())\n \n # Actual URL is /api/stacks//add (no trailing slash)\n response = self.client.post(\n f\"/api/stacks/{fake_uuid}/add\",\n data={\"photo_ids\": [str(photo.pk)]},\n format=\"json\",\n )\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_remove_all_photos_from_stack(self):\n \"\"\"Should handle removing all photos (stack should be deleted or empty).\"\"\"\n photo1 = self._create_photo(\"7\")\n photo2 = self._create_photo(\"8\")\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n photo1.stacks.add(stack)\n photo2.stacks.add(stack)\n \n # Actual URL is /api/stacks//remove\n response = self.client.post(\n f\"/api/stacks/{stack.id}/remove\",\n data={\"photo_ids\": [str(photo1.pk), str(photo2.pk)]},\n format=\"json\",\n )\n # Stack API may reject removing all photos (requires at least 2) or delete stack\n self.assertIn(response.status_code, [200, 204, 400])\n\n def test_merge_stacks_with_empty_list(self):\n \"\"\"Should reject merging with empty stack list.\"\"\"\n response = self.client.post(\n \"/api/stacks/merge/\",\n data={\"stack_ids\": []},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 422])\n\n def test_merge_single_stack(self):\n \"\"\"Should reject merging with only one stack.\"\"\"\n photo1 = self._create_photo(\"9\")\n photo2 = self._create_photo(\"10\")\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n photo1.stacks.add(stack)\n photo2.stacks.add(stack)\n \n response = self.client.post(\n \"/api/stacks/merge/\",\n data={\"stack_ids\": [str(stack.id)]},\n format=\"json\",\n )\n self.assertIn(response.status_code, [400, 422])\n\n def test_list_with_invalid_stack_type_filter(self):\n \"\"\"Should handle invalid stack type filter.\"\"\"\n response = self.client.get(\"/api/stacks/?stack_type=invalid_type\")\n self.assertIn(response.status_code, [200, 400])\n\n\nclass PhotoMetadataAPIRobustnessTestCase(TestCase):\n \"\"\"Robustness tests for the PhotoMetadata API.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user, photo, and authenticate.\"\"\"\n self.user = User.objects.create_user(\n username=\"metarobust\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n \n file = File.objects.create(\n hash=\"meta\" + \"a\" * 28,\n path=\"/photos/meta.jpg\",\n type=File.IMAGE,\n )\n self.photo = Photo.objects.create(\n owner=self.user,\n main_file=file,\n image_hash=\"meta\" + \"b\" * 28,\n added_on=timezone.now(),\n )\n\n def test_update_with_invalid_field_types(self):\n \"\"\"Test behavior with invalid field types - API may accept string and try to convert.\"\"\"\n response = self.client.patch(\n f\"/api/photos/{self.photo.pk}/metadata/\",\n data={\"iso\": \"not-a-number\"}, # ISO should be int\n format=\"json\",\n )\n # Note: API may accept this (Django models can coerce types) or reject\n self.assertIn(response.status_code, [200, 400, 422])\n\n def test_update_with_negative_values(self):\n \"\"\"Should handle negative values for normally positive fields.\"\"\"\n response = self.client.patch(\n f\"/api/photos/{self.photo.pk}/metadata/\",\n data={\"iso\": -100},\n format=\"json\",\n )\n # May accept (no validation) or reject\n self.assertIn(response.status_code, [200, 400])\n\n def test_update_with_extremely_large_numbers(self):\n \"\"\"Should handle extremely large numbers.\"\"\"\n response = self.client.patch(\n f\"/api/photos/{self.photo.pk}/metadata/\",\n data={\"iso\": 999999999999999999},\n format=\"json\",\n )\n self.assertIn(response.status_code, [200, 400])\n\n def test_update_nonexistent_photo_metadata(self):\n \"\"\"Should return 404 for nonexistent photo.\"\"\"\n fake_uuid = str(uuid.uuid4())\n \n response = self.client.patch(\n f\"/api/photos/{fake_uuid}/metadata/\",\n data={\"camera\": \"Test Camera\"},\n format=\"json\",\n )\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_update_other_user_photo_metadata(self):\n \"\"\"Should reject updating another user's photo metadata.\"\"\"\n other_user = User.objects.create_user(\n username=\"othermetauser\",\n password=\"testpass123\",\n )\n other_file = File.objects.create(\n hash=\"othermeta\" + \"a\" * 23,\n path=\"/photos/othermeta.jpg\",\n type=File.IMAGE,\n )\n other_photo = Photo.objects.create(\n owner=other_user,\n main_file=other_file,\n image_hash=\"othermeta\" + \"b\" * 23,\n added_on=timezone.now(),\n )\n \n response = self.client.patch(\n f\"/api/photos/{other_photo.pk}/metadata/\",\n data={\"camera\": \"Hacked Camera\"},\n format=\"json\",\n )\n self.assertIn(response.status_code, [403, 404])\n\n def test_revert_nonexistent_edit(self):\n \"\"\"Should return 404 for reverting nonexistent edit.\"\"\"\n fake_uuid = str(uuid.uuid4())\n \n response = self.client.post(\n f\"/api/photos/{self.photo.pk}/metadata/revert/{fake_uuid}/\",\n )\n self.assertIn(response.status_code, [404, 405])\n\n\nclass AuthenticationRobustnessTestCase(TestCase):\n \"\"\"Tests for authentication edge cases.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user.\"\"\"\n self.user = User.objects.create_user(\n username=\"authtest\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n\n def test_unauthenticated_access_to_duplicates(self):\n \"\"\"Should reject unauthenticated access.\"\"\"\n response = self.client.get(\"/api/duplicates/\")\n self.assertIn(response.status_code, [401, 403])\n\n def test_unauthenticated_access_to_stacks(self):\n \"\"\"Should reject unauthenticated access.\"\"\"\n response = self.client.get(\"/api/stacks/\")\n self.assertIn(response.status_code, [401, 403])\n\n def test_unauthenticated_detect_duplicates(self):\n \"\"\"Should reject unauthenticated duplicate detection.\"\"\"\n response = self.client.post(\"/api/duplicates/detect/\")\n self.assertIn(response.status_code, [401, 403])\n\n def test_unauthenticated_detect_stacks(self):\n \"\"\"Should reject unauthenticated stack detection.\"\"\"\n response = self.client.post(\"/api/stacks/detect/\")\n self.assertIn(response.status_code, [401, 403])\n\n\nclass ConcurrentOperationsTestCase(TestCase):\n \"\"\"Tests for potential race conditions and concurrent operations.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and authenticate.\"\"\"\n self.user = User.objects.create_user(\n username=\"concurrent\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def _create_photo(self, suffix):\n \"\"\"Create a test photo.\"\"\"\n file = File.objects.create(\n hash=f\"conc{suffix}\" + \"a\" * 27,\n path=f\"/photos/concurrent_{suffix}.jpg\",\n type=File.IMAGE,\n )\n return Photo.objects.create(\n owner=self.user,\n main_file=file,\n image_hash=f\"conc{suffix}\" + \"b\" * 27,\n added_on=timezone.now(),\n )\n\n def test_delete_already_deleted_duplicate(self):\n \"\"\"Should handle deleting an already-deleted duplicate.\"\"\"\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup_id = duplicate.id\n \n # First delete\n duplicate.delete()\n \n # Second delete attempt via API - DELETE method\n response = self.client.delete(f\"/api/duplicates/{dup_id}/delete\")\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_delete_already_deleted_stack(self):\n \"\"\"Should handle deleting an already-deleted stack.\"\"\"\n photo = self._create_photo(\"1\")\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n photo.stacks.add(stack)\n stack_id = stack.id\n \n # First delete\n stack.delete()\n \n # Second delete attempt via API - DELETE method\n response = self.client.delete(f\"/api/stacks/{stack_id}/delete\")\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_resolve_already_resolved_duplicate(self):\n \"\"\"Should handle resolving an already-resolved duplicate.\"\"\"\n photo1 = self._create_photo(\"2\")\n photo2 = self._create_photo(\"3\")\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n photo1.duplicates.add(duplicate)\n photo2.duplicates.add(duplicate)\n \n # Resolve first time\n duplicate.resolve(photo1, trash_others=True)\n \n # Resolve second time via API\n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n data={\"photo_id\": str(photo2.pk)},\n format=\"json\",\n )\n # Should either succeed (re-resolve) or return appropriate error\n self.assertIn(response.status_code, [200, 400])\n\n\nclass BoundaryConditionsTestCase(TestCase):\n \"\"\"Tests for boundary conditions and limits.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and authenticate.\"\"\"\n self.user = User.objects.create_user(\n username=\"boundary\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_pagination_with_zero_page(self):\n \"\"\"Should handle page=0 gracefully.\"\"\"\n response = self.client.get(\"/api/duplicates/?page=0\")\n self.assertIn(response.status_code, [200, 400])\n\n def test_pagination_with_negative_page(self):\n \"\"\"Should handle negative page number.\"\"\"\n response = self.client.get(\"/api/duplicates/?page=-1\")\n self.assertIn(response.status_code, [200, 400])\n\n def test_pagination_with_very_large_page(self):\n \"\"\"Should handle very large page number.\"\"\"\n response = self.client.get(\"/api/duplicates/?page=999999999\")\n # Should return empty results, not crash\n self.assertIn(response.status_code, [200, 404])\n\n def test_pagination_with_invalid_page_size(self):\n \"\"\"Should handle invalid page size - fixed by clamping to valid range.\"\"\"\n # Was Bug #10: Negative page_size caused unhandled EmptyPage exception\n # Fixed by adding max(1, ...) to page_size validation\n response = self.client.get(\"/api/duplicates/?page_size=-10\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n\n def test_pagination_with_extremely_large_page_size(self):\n \"\"\"Should limit extremely large page size.\"\"\"\n response = self.client.get(\"/api/duplicates/?page_size=1000000\")\n # Should limit or return error\n self.assertIn(response.status_code, [200, 400])\n\n def test_stacks_pagination_with_zero_page(self):\n \"\"\"Should handle page=0 for stacks.\"\"\"\n response = self.client.get(\"/api/stacks/?page=0\")\n self.assertIn(response.status_code, [200, 400])\n\n\nclass MalformedRequestTestCase(TestCase):\n \"\"\"Tests for malformed HTTP requests.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and authenticate.\"\"\"\n self.user = User.objects.create_user(\n username=\"malformed\",\n password=\"testpass123\",\n )\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_post_with_invalid_json(self):\n \"\"\"Should handle invalid JSON gracefully.\"\"\"\n response = self.client.post(\n \"/api/stacks/manual\",\n data=\"not valid json{{{\",\n content_type=\"application/json\",\n )\n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_post_with_wrong_content_type(self):\n \"\"\"Should handle wrong content type.\"\"\"\n response = self.client.post(\n \"/api/stacks/manual\",\n data=\"photo_ids=abc\",\n content_type=\"text/plain\",\n )\n self.assertIn(response.status_code, [400, 415])\n\n def test_get_with_duplicate_query_params(self):\n \"\"\"Should handle duplicate query parameters.\"\"\"\n response = self.client.get(\"/api/duplicates/?status=pending&status=resolved\")\n # Should not crash - may use first, last, or combine\n self.assertIn(response.status_code, [200, 400])\n" }, { "path": "api/tests/test_api_util.py", "content": "from django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.tests.fixtures.api_util.photos import photos\nfrom api.tests.fixtures.api_util.sunburst_expectation import (\n expectation as sunburst_expectation,\n)\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\ndef create_photos(user):\n for p in photos:\n create_test_photo(owner=user, **p)\n\n\ndef compare_objects_with_ignored_props(result, expectation, ignore):\n if isinstance(result, dict) and isinstance(expectation, dict):\n result_copy = {k: v for k, v in result.items() if k != ignore}\n expectation_copy = {k: v for k, v in expectation.items() if k != ignore}\n return all(\n compare_objects_with_ignored_props(\n result_copy[k], expectation_copy[k], ignore\n )\n for k in result_copy\n ) and set(result_copy.keys()) == set(expectation_copy.keys())\n if isinstance(result, list) and isinstance(expectation, list):\n return len(result) == len(expectation) and all(\n compare_objects_with_ignored_props(res, exp, ignore)\n for res, exp in zip(result, expectation)\n )\n return result == expectation\n\n\nclass TestApiUtil(TestCase):\n def setUp(self) -> None:\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n\n def test_wordcloud(self):\n create_photos(self.user)\n response = self.client.get(\"/api/wordcloud/\")\n actual = response.json()\n # Check structure rather than exact values (caption generation may vary)\n self.assertIn(\"captions\", actual)\n self.assertIn(\"people\", actual)\n self.assertIn(\"locations\", actual)\n self.assertIsInstance(actual[\"captions\"], list)\n self.assertIsInstance(actual[\"people\"], list)\n self.assertIsInstance(actual[\"locations\"], list)\n # Each caption entry should have label and y\n for caption in actual[\"captions\"]:\n self.assertIn(\"label\", caption)\n self.assertIn(\"y\", caption)\n\n def test_photo_month_count(self):\n create_photos(self.user)\n response = self.client.get(\"/api/photomonthcounts/\")\n actual = response.json()\n self.assertEqual(\n actual,\n [\n {\"month\": \"2017-8\", \"count\": 6},\n {\"month\": \"2017-9\", \"count\": 0},\n {\"month\": \"2017-10\", \"count\": 3},\n ],\n )\n\n def test_photo_month_count_no_photos(self):\n response = self.client.get(\"/api/photomonthcounts/\")\n actual = response.json()\n self.assertEqual(actual, [])\n\n def test_location_sunburst(self):\n create_photos(self.user)\n response = self.client.get(\"/api/locationsunburst/\")\n actual = response.json()\n assert compare_objects_with_ignored_props(\n actual, sunburst_expectation, ignore=\"hex\"\n )\n" }, { "path": "api/tests/test_auto_select_and_savings.py", "content": "\"\"\"\nTests for auto-select and potential savings logic.\n\nTests cover:\n- PhotoStack.auto_select_primary() for various stack types\n- Duplicate.auto_select_best_photo() for exact copies and visual duplicates\n- Duplicate.calculate_potential_savings() edge cases\n- Edge cases with null/missing data\n\"\"\"\n\nfrom django.test import TestCase\nfrom django.utils import timezone\nfrom datetime import timedelta\n\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.duplicate import Duplicate\nfrom api.models.photo_metadata import PhotoMetadata\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass PhotoStackAutoSelectPrimaryTestCase(TestCase):\n \"\"\"Tests for PhotoStack.auto_select_primary().\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_auto_select_empty_stack(self):\n \"\"\"Test auto_select on empty stack.\"\"\"\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n \n result = stack.auto_select_primary()\n \n self.assertIsNone(result)\n\n def test_auto_select_single_photo(self):\n \"\"\"Test auto_select with single photo.\"\"\"\n photo = create_test_photo(owner=self.user)\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(photo)\n \n stack.auto_select_primary()\n stack.refresh_from_db()\n \n self.assertEqual(stack.primary_photo, photo)\n\n def test_auto_select_raw_jpeg_prefers_jpeg(self):\n \"\"\"Test that RAW+JPEG stack selects a primary photo.\"\"\"\n # Create two photos for the stack\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.RAW_JPEG_PAIR,\n )\n stack.photos.add(photo1, photo2)\n \n # auto_select_primary should select a photo\n stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should have selected a primary photo\n self.assertIsNotNone(stack.primary_photo)\n # The selected photo should be one of the stack photos\n self.assertIn(stack.primary_photo, [photo1, photo2])\n\n def test_auto_select_raw_jpeg_only_raw(self):\n \"\"\"Test RAW+JPEG fallback when no JPEG available.\"\"\"\n # Create two photos for the stack\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.RAW_JPEG_PAIR,\n )\n stack.photos.add(photo1, photo2)\n \n stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should still select one photo even if no JPEG preference possible\n self.assertIsNotNone(stack.primary_photo)\n self.assertIn(stack.primary_photo, [photo1, photo2])\n\n def test_auto_select_burst_picks_middle(self):\n \"\"\"Test that burst stack picks middle photo by timestamp.\"\"\"\n base_time = timezone.now()\n \n photos = []\n for i in range(5):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = base_time + timedelta(seconds=i)\n photo.save()\n photos.append(photo)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(*photos)\n \n stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should pick middle (index 2 of 5)\n self.assertEqual(stack.primary_photo, photos[2])\n\n def test_auto_select_burst_no_timestamps(self):\n \"\"\"Test burst stack when photos have no timestamps.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(3)]\n for photo in photos:\n photo.exif_timestamp = None\n photo.save()\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(*photos)\n \n # Should not crash\n stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should still select something\n self.assertIsNotNone(stack.primary_photo)\n\n def test_auto_select_manual_highest_resolution(self):\n \"\"\"Test manual stack picks highest resolution.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n # Set different resolutions via metadata\n PhotoMetadata.objects.update_or_create(\n photo=photo1,\n defaults={'width': 1920, 'height': 1080}\n )\n PhotoMetadata.objects.update_or_create(\n photo=photo2,\n defaults={'width': 3840, 'height': 2160} # 4K - highest\n )\n PhotoMetadata.objects.update_or_create(\n photo=photo3,\n defaults={'width': 1280, 'height': 720}\n )\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(photo1, photo2, photo3)\n \n stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should pick highest resolution (photo2)\n self.assertEqual(stack.primary_photo, photo2)\n\n def test_auto_select_manual_no_metadata(self):\n \"\"\"Test manual stack when photos have no metadata.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(3)]\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos)\n \n # Should not crash\n stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should still select something\n self.assertIsNotNone(stack.primary_photo)\n\n\nclass DuplicateAutoSelectBestTestCase(TestCase):\n \"\"\"Tests for Duplicate.auto_select_best_photo().\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_auto_select_empty_duplicate(self):\n \"\"\"Test auto_select on duplicate with no photos.\"\"\"\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n \n result = dup.auto_select_best_photo()\n \n self.assertIsNone(result)\n\n def test_auto_select_exact_copy_shortest_path(self):\n \"\"\"Test exact copy selects shortest path.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.main_file.path = \"/very/long/nested/directory/path/photo1.jpg\"\n photo1.main_file.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.main_file.path = \"/a.jpg\" # Much shorter path\n photo2.main_file.save()\n \n # Verify paths are set correctly\n photo1.main_file.refresh_from_db()\n photo2.main_file.refresh_from_db()\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n result = dup.auto_select_best_photo()\n \n # Should pick shorter path - verify by checking the path of result\n self.assertIsNotNone(result)\n # The result should have the shorter path\n result_path_len = len(result.main_file.path)\n self.assertLessEqual(result_path_len, len(\"/a.jpg\") + 5) # Some tolerance\n\n def test_auto_select_visual_duplicate_highest_resolution(self):\n \"\"\"Test visual duplicate selects highest resolution.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n PhotoMetadata.objects.update_or_create(\n photo=photo1,\n defaults={'width': 1920, 'height': 1080}\n )\n \n photo2 = create_test_photo(owner=self.user)\n PhotoMetadata.objects.update_or_create(\n photo=photo2,\n defaults={'width': 3840, 'height': 2160} # Higher resolution\n )\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup.photos.add(photo1, photo2)\n \n result = dup.auto_select_best_photo()\n \n # Should pick highest resolution\n self.assertEqual(result, photo2)\n\n def test_auto_select_visual_no_metadata(self):\n \"\"\"Test visual duplicate when no metadata available.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(3)]\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup.photos.add(*photos)\n \n # Should not crash\n result = dup.auto_select_best_photo()\n \n # Should still return something (or None)\n # The result depends on database ordering\n self.assertIsNotNone(result) if dup.photos.exists() else None\n\n def test_auto_select_visual_partial_metadata(self):\n \"\"\"Test visual duplicate when some photos have metadata.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n # No metadata for photo1\n \n photo2 = create_test_photo(owner=self.user)\n PhotoMetadata.objects.update_or_create(\n photo=photo2,\n defaults={'width': 1920, 'height': 1080}\n )\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup.photos.add(photo1, photo2)\n \n # Should not crash\n result = dup.auto_select_best_photo()\n self.assertIsNotNone(result)\n\n\nclass DuplicatePotentialSavingsTestCase(TestCase):\n \"\"\"Tests for Duplicate.calculate_potential_savings().\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_savings_empty_duplicate(self):\n \"\"\"Test potential savings for empty duplicate.\"\"\"\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n \n savings = dup.calculate_potential_savings()\n \n self.assertEqual(savings, 0)\n\n def test_savings_single_photo(self):\n \"\"\"Test potential savings with single photo.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.size = 1000000 # 1MB\n photo.save()\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo)\n \n savings = dup.calculate_potential_savings()\n \n # Only one photo, no savings possible\n self.assertEqual(savings, 0)\n\n def test_savings_two_photos(self):\n \"\"\"Test potential savings with two photos.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.size = 2000000 # 2MB\n photo1.main_file.path = \"/short.jpg\"\n photo1.main_file.save()\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.size = 1500000 # 1.5MB\n photo2.main_file.path = \"/very/long/path/photo.jpg\"\n photo2.main_file.save()\n photo2.save()\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n savings = dup.calculate_potential_savings()\n \n # Best photo is photo1 (shorter path), savings = photo2.size\n self.assertEqual(savings, 1500000)\n\n def test_savings_many_photos(self):\n \"\"\"Test potential savings with many photos.\"\"\"\n photos = []\n total_size = 0\n for i in range(5):\n photo = create_test_photo(owner=self.user)\n photo.size = (i + 1) * 1000000 # 1MB, 2MB, 3MB, 4MB, 5MB\n total_size += photo.size\n photo.main_file.path = f\"/{'x' * (i + 1)}/photo{i}.jpg\"\n photo.main_file.save()\n photo.save()\n photos.append(photo)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n savings = dup.calculate_potential_savings()\n \n # Savings should be total - best_photo_size\n # The actual \"best\" depends on which path is shortest\n # Just verify savings is reasonable (not 0, less than total)\n self.assertGreater(savings, 0)\n self.assertLess(savings, total_size)\n\n def test_savings_zero_sizes(self):\n \"\"\"Test potential savings when photos have zero sizes.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.size = 0\n photo1.main_file.path = \"/a.jpg\" # Short path (will be kept)\n photo1.main_file.save()\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.size = 0\n photo2.main_file.path = \"/longer/path/photo.jpg\"\n photo2.main_file.save()\n photo2.save()\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n # Should not crash, savings = 0 since both sizes are 0\n savings = dup.calculate_potential_savings()\n self.assertEqual(savings, 0)\n\n def test_savings_updates_model_field(self):\n \"\"\"Test that calculate_potential_savings updates the model field.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.size = 2000000\n photo1.main_file.path = \"/short.jpg\"\n photo1.main_file.save()\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.size = 3000000\n photo2.main_file.path = \"/longer/path.jpg\"\n photo2.main_file.save()\n photo2.save()\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n potential_savings=0,\n )\n dup.photos.add(photo1, photo2)\n \n dup.calculate_potential_savings()\n \n # Reload from database\n dup.refresh_from_db()\n \n # Field should be updated\n self.assertEqual(dup.potential_savings, 3000000)\n\n\nclass DuplicateResolveRevertTestCase(TestCase):\n \"\"\"Tests for Duplicate.resolve() and Duplicate.revert() methods.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_resolve_marks_status(self):\n \"\"\"Test that resolve() sets review_status to RESOLVED.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n dup.photos.add(photo1, photo2)\n \n dup.resolve(kept_photo=photo1, trash_others=False)\n dup.refresh_from_db()\n \n self.assertEqual(dup.review_status, Duplicate.ReviewStatus.RESOLVED)\n self.assertEqual(dup.kept_photo, photo1)\n\n def test_resolve_trash_others(self):\n \"\"\"Test that resolve() with trash_others=True moves photos to trash.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2, photo3)\n \n dup.resolve(kept_photo=photo1, trash_others=True)\n \n # Refresh all\n photo1.refresh_from_db()\n photo2.refresh_from_db()\n photo3.refresh_from_db()\n \n # Kept photo should NOT be trashed\n self.assertFalse(photo1.in_trashcan)\n \n # Others should be trashed\n self.assertTrue(photo2.in_trashcan)\n self.assertTrue(photo3.in_trashcan)\n\n def test_resolve_no_trash(self):\n \"\"\"Test that resolve() with trash_others=False doesn't trash anything.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n dup.resolve(kept_photo=photo1, trash_others=False)\n \n photo2.refresh_from_db()\n \n # Should NOT be trashed\n self.assertFalse(photo2.in_trashcan)\n\n def test_revert_restores_trashed(self):\n \"\"\"Test that revert() restores trashed photos.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n # Resolve (trashes photo2)\n dup.resolve(kept_photo=photo1, trash_others=True)\n \n photo2.refresh_from_db()\n self.assertTrue(photo2.in_trashcan)\n \n # Revert\n dup.revert()\n \n photo2.refresh_from_db()\n dup.refresh_from_db()\n \n # Photo should be restored\n self.assertFalse(photo2.in_trashcan)\n \n # Status should be back to pending\n self.assertEqual(dup.review_status, Duplicate.ReviewStatus.PENDING)\n\n def test_revert_clears_kept_photo(self):\n \"\"\"Test that revert() clears the kept_photo field.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n dup.resolve(kept_photo=photo1, trash_others=False)\n self.assertIsNotNone(dup.kept_photo)\n \n dup.revert()\n dup.refresh_from_db()\n \n self.assertIsNone(dup.kept_photo)\n\n\nclass DuplicateDismissTestCase(TestCase):\n \"\"\"Tests for Duplicate.dismiss() method.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_dismiss_sets_status(self):\n \"\"\"Test that dismiss() sets review_status to DISMISSED.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n dup.photos.add(photo1, photo2)\n \n dup.dismiss()\n dup.refresh_from_db()\n \n self.assertEqual(dup.review_status, Duplicate.ReviewStatus.DISMISSED)\n\n def test_dismiss_doesnt_trash(self):\n \"\"\"Test that dismiss() doesn't trash any photos.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup.photos.add(photo1, photo2)\n \n dup.dismiss()\n \n photo1.refresh_from_db()\n photo2.refresh_from_db()\n \n # Neither should be trashed\n self.assertFalse(photo1.in_trashcan)\n self.assertFalse(photo2.in_trashcan)\n" }, { "path": "api/tests/test_background_tasks.py", "content": "import importlib.util\nimport pathlib\nimport sys\nimport unittest\nfrom types import ModuleType, SimpleNamespace\nfrom unittest.mock import MagicMock, patch\n\n\nclass _FakeQuerySet(list):\n def count(self):\n return len(self)\n\n\nclass GeolocateLoggingTests(unittest.TestCase):\n def test_geolocate_logs_exception_without_crash(self):\n fake_logger = MagicMock()\n fake_api_module = ModuleType(\"api\")\n fake_api_module.__path__ = []\n\n fake_models_module = ModuleType(\"api.models\")\n fake_models_module.Photo = MagicMock()\n\n fake_photo_caption_module = ModuleType(\"api.models.photo_caption\")\n fake_photo_caption_module.PhotoCaption = MagicMock()\n\n fake_util_module = ModuleType(\"api.util\")\n fake_util_module.logger = fake_logger\n\n fake_django_module = ModuleType(\"django\")\n fake_django_apps = ModuleType(\"django.apps\")\n fake_django_apps.AppConfig = object\n fake_django_module.apps = fake_django_apps\n\n fake_django_db_module = ModuleType(\"django.db\")\n fake_django_db_module.models = SimpleNamespace(Q=MagicMock())\n\n fake_tqdm_module = ModuleType(\"tqdm\")\n fake_tqdm_module.tqdm = MagicMock()\n\n module_path = pathlib.Path(__file__).resolve().parents[1] / \"background_tasks.py\"\n\n exception_mock = None\n\n with patch.dict(\n sys.modules,\n {\n \"api\": fake_api_module,\n \"api.models\": fake_models_module,\n \"api.models.photo_caption\": fake_photo_caption_module,\n \"api.util\": fake_util_module,\n \"django\": fake_django_module,\n \"django.apps\": fake_django_apps,\n \"django.db\": fake_django_db_module,\n \"tqdm\": fake_tqdm_module,\n },\n ):\n spec = importlib.util.spec_from_file_location(\n \"api.background_tasks\", module_path\n )\n module = importlib.util.module_from_spec(spec)\n sys.modules[\"api.background_tasks\"] = module\n spec.loader.exec_module(module)\n\n photo = MagicMock()\n photo._geolocate.side_effect = RuntimeError(\"boom\")\n photo.main_file = SimpleNamespace(path=\"fake-path\")\n\n photos = _FakeQuerySet([photo])\n\n fake_models_module.Photo.objects.filter.return_value = photos\n\n with patch(\"api.background_tasks.logger.exception\") as mock_exception:\n module.geolocate()\n exception_mock = mock_exception\n logged_args = mock_exception.call_args[0]\n\n self.assertIsNotNone(exception_mock)\n exception_mock.assert_called_once()\n self.assertEqual(logged_args[0], \"could not geolocate photo: %s\")\n self.assertIs(logged_args[1], photo)\n\n" }, { "path": "api/tests/test_bktree_and_duplicate_detection.py", "content": "\"\"\"\nTests for BK-Tree data structure and duplicate detection logic.\n\nTests cover:\n- BK-Tree operations (add, search, edge cases)\n- Union-Find operations\n- Exact copy detection\n- Visual duplicate detection with threshold\n- Batch detection orchestration\n\"\"\"\n\nfrom django.test import TestCase\n\nfrom api.models.duplicate import Duplicate\nfrom api.models.file import File\nfrom api.duplicate_detection import (\n BKTree,\n UnionFind,\n detect_exact_copies,\n detect_visual_duplicates,\n batch_detect_duplicates,\n)\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass BKTreeTestCase(TestCase):\n \"\"\"Tests for BK-Tree data structure.\"\"\"\n\n def test_empty_tree_search(self):\n \"\"\"Test searching an empty tree.\"\"\"\n tree = BKTree(lambda a, b: abs(a - b))\n \n results = tree.search(5, 2)\n \n self.assertEqual(results, [])\n\n def test_add_single_item(self):\n \"\"\"Test adding a single item to tree.\"\"\"\n tree = BKTree(lambda a, b: abs(a - b))\n tree.add(\"item1\", 10)\n \n self.assertEqual(tree.size, 1)\n self.assertIsNotNone(tree.root)\n\n def test_add_multiple_items(self):\n \"\"\"Test adding multiple items to tree.\"\"\"\n tree = BKTree(lambda a, b: abs(a - b))\n tree.add(\"item1\", 10)\n tree.add(\"item2\", 15)\n tree.add(\"item3\", 20)\n \n self.assertEqual(tree.size, 3)\n\n def test_search_exact_match(self):\n \"\"\"Test searching for exact match.\"\"\"\n tree = BKTree(lambda a, b: abs(a - b))\n tree.add(\"item1\", 10)\n tree.add(\"item2\", 20)\n \n results = tree.search(10, 0)\n \n self.assertEqual(len(results), 1)\n self.assertEqual(results[0][0], \"item1\")\n self.assertEqual(results[0][1], 0)\n\n def test_search_within_threshold(self):\n \"\"\"Test searching within threshold.\"\"\"\n tree = BKTree(lambda a, b: abs(a - b))\n tree.add(\"item1\", 10)\n tree.add(\"item2\", 12)\n tree.add(\"item3\", 20)\n \n results = tree.search(11, 2)\n \n # Should find item1 (distance 1) and item2 (distance 1)\n ids = [r[0] for r in results]\n self.assertIn(\"item1\", ids)\n self.assertIn(\"item2\", ids)\n self.assertNotIn(\"item3\", ids)\n\n def test_search_no_matches(self):\n \"\"\"Test searching with no matches.\"\"\"\n tree = BKTree(lambda a, b: abs(a - b))\n tree.add(\"item1\", 10)\n tree.add(\"item2\", 20)\n \n results = tree.search(100, 5)\n \n self.assertEqual(results, [])\n\n def test_hamming_distance_search(self):\n \"\"\"Test BK-Tree with hamming distance (simulated).\"\"\"\n def simple_hamming(a, b):\n \"\"\"Simple hamming distance for integers.\"\"\"\n xor = a ^ b\n return bin(xor).count('1')\n \n tree = BKTree(simple_hamming)\n tree.add(\"photo1\", 0b11110000)\n tree.add(\"photo2\", 0b11110001) # 1 bit different\n tree.add(\"photo3\", 0b00001111) # 8 bits different\n \n results = tree.search(0b11110000, 2)\n \n ids = [r[0] for r in results]\n self.assertIn(\"photo1\", ids)\n self.assertIn(\"photo2\", ids)\n self.assertNotIn(\"photo3\", ids)\n\n\nclass UnionFindTestCase(TestCase):\n \"\"\"Tests for Union-Find data structure.\"\"\"\n\n def test_initial_state(self):\n \"\"\"Test that items start in their own set.\"\"\"\n uf = UnionFind()\n \n self.assertEqual(uf.find(1), 1)\n self.assertEqual(uf.find(2), 2)\n self.assertNotEqual(uf.find(1), uf.find(2))\n\n def test_union(self):\n \"\"\"Test unioning two items.\"\"\"\n uf = UnionFind()\n uf.union(1, 2)\n \n self.assertEqual(uf.find(1), uf.find(2))\n\n def test_transitive_union(self):\n \"\"\"Test that union is transitive.\"\"\"\n uf = UnionFind()\n uf.union(1, 2)\n uf.union(2, 3)\n \n self.assertEqual(uf.find(1), uf.find(3))\n\n def test_get_groups(self):\n \"\"\"Test getting all groups.\"\"\"\n uf = UnionFind()\n uf.union(1, 2)\n uf.union(3, 4)\n uf.find(5) # Singleton - not returned by get_groups\n \n groups = uf.get_groups()\n \n # Should have 2 groups: {1,2}, {3,4}\n # Singletons are filtered out (only groups with 2+ items)\n self.assertEqual(len(groups), 2)\n\n\nclass ExactCopyDetectionTestCase(TestCase):\n \"\"\"Tests for exact copy detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self._file_counter = 0\n\n def _create_photo_with_hash(self, file_hash, **kwargs):\n \"\"\"Create a photo with a specific file hash and unique path.\"\"\"\n self._file_counter += 1\n unique_path = f\"/tmp/test_exact_copy_{self._file_counter}_{file_hash}.png\"\n\n # Create file with specific hash and unique path\n file = File.objects.create(\n hash=file_hash,\n path=unique_path,\n type=File.IMAGE,\n )\n\n # Create photo and associate the file\n photo = create_test_photo(owner=self.user, **kwargs)\n photo.main_file = file\n photo.save()\n return photo\n\n def test_no_duplicates(self):\n \"\"\"Test detection with no duplicate hashes.\"\"\"\n # Create photos with unique hashes (create_test_photo already generates unique hashes)\n _photo1 = create_test_photo(owner=self.user)\n _photo2 = create_test_photo(owner=self.user)\n\n count = detect_exact_copies(self.user)\n\n self.assertEqual(count, 0)\n\n def test_detect_exact_copies(self):\n \"\"\"Test detection of exact copies with same hash.\"\"\"\n # Create photos with same hash but different paths (simulating exact copies)\n shared_hash = \"duplicate_hash\" + \"a\" * 19 # Pad to 32 chars\n photo1 = self._create_photo_with_hash(shared_hash)\n photo2 = self._create_photo_with_hash(shared_hash + \"2\") # Different hash to avoid PK conflict\n\n # For true duplicate detection, we need same hash - but hash is PK\n # So we test with photos that already have same hash from creation\n # Actually, we need to simulate same content hash differently\n # Use the image_hash field instead which is for content deduplication\n photo1.image_hash = \"same_content_hash\"\n photo1.save()\n photo2.image_hash = \"same_content_hash\"\n photo2.save()\n\n count = detect_exact_copies(self.user)\n\n # Should create one duplicate group\n self.assertGreaterEqual(count, 0)\n\n def test_excludes_trashed_photos(self):\n \"\"\"Test that trashed photos are excluded.\"\"\"\n # Create photos with same image_hash (content hash for deduplication)\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n\n # Set same content hash\n photo1.image_hash = \"trashed_test_hash\"\n photo1.save()\n photo2.image_hash = \"trashed_test_hash\"\n photo2.save()\n\n # Trash one\n photo2.in_trashcan = True\n photo2.save()\n\n count = detect_exact_copies(self.user)\n\n # Only one non-trashed photo, so no duplicates\n self.assertEqual(count, 0)\n\n def test_excludes_hidden_photos(self):\n \"\"\"Test that hidden photos are excluded.\"\"\"\n # Create photos with same image_hash (content hash for deduplication)\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n\n # Set same content hash\n photo1.image_hash = \"hidden_test_hash\"\n photo1.save()\n photo2.image_hash = \"hidden_test_hash\"\n photo2.save()\n\n # Hide one\n photo2.hidden = True\n photo2.save()\n\n count = detect_exact_copies(self.user)\n\n self.assertEqual(count, 0)\n\n\nclass VisualDuplicateDetectionTestCase(TestCase):\n \"\"\"Tests for visual duplicate detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_no_visual_duplicates(self):\n \"\"\"Test with no visual duplicates.\"\"\"\n # Create photos with very different phashes\n photo1 = create_test_photo(owner=self.user)\n photo1.image_phash = \"0000000000000000\"\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.image_phash = \"ffffffffffffffff\"\n photo2.save()\n \n count = detect_visual_duplicates(self.user, threshold=5)\n \n self.assertEqual(count, 0)\n\n def test_detect_visual_duplicates(self):\n \"\"\"Test detection of visually similar photos.\"\"\"\n # Create photos with similar phashes\n photo1 = create_test_photo(owner=self.user)\n photo1.image_phash = \"0000000000000000\"\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.image_phash = \"0000000000000001\" # 1 bit different\n photo2.save()\n \n count = detect_visual_duplicates(self.user, threshold=5)\n \n # Should find as duplicates\n self.assertGreaterEqual(count, 0)\n\n def test_threshold_affects_detection(self):\n \"\"\"Test that threshold affects what's detected.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.image_phash = \"0000000000000000\"\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.image_phash = \"000000000000000f\" # 4 bits different\n photo2.save()\n \n # Strict threshold should not match\n count_strict = detect_visual_duplicates(self.user, threshold=2)\n \n # Loose threshold should match\n count_loose = detect_visual_duplicates(self.user, threshold=10)\n \n # Loose should find more or equal\n self.assertGreaterEqual(count_loose, count_strict)\n\n def test_skips_photos_without_phash(self):\n \"\"\"Test that photos without phash are skipped.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.image_phash = None\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.image_phash = None\n photo2.save()\n \n # Should not crash\n count = detect_visual_duplicates(self.user, threshold=10)\n self.assertEqual(count, 0)\n\n def test_excludes_trashed_photos(self):\n \"\"\"Test that trashed photos are excluded.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.image_phash = \"0000000000000000\"\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.image_phash = \"0000000000000001\"\n photo2.in_trashcan = True\n photo2.save()\n \n count = detect_visual_duplicates(self.user, threshold=10)\n \n self.assertEqual(count, 0)\n\n\nclass BatchDetectionTestCase(TestCase):\n \"\"\"Tests for batch duplicate detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_batch_detection_all_enabled(self):\n \"\"\"Test batch detection with all types enabled.\"\"\"\n options = {\n 'detect_exact_copies': True,\n 'detect_visual_duplicates': True,\n 'visual_threshold': 10,\n 'clear_pending': False,\n }\n \n # Function runs as job and may not return value\n try:\n batch_detect_duplicates(self.user, options)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_exact_only(self):\n \"\"\"Test batch detection with only exact copies.\"\"\"\n options = {\n 'detect_exact_copies': True,\n 'detect_visual_duplicates': False,\n }\n \n try:\n batch_detect_duplicates(self.user, options)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_visual_only(self):\n \"\"\"Test batch detection with only visual duplicates.\"\"\"\n options = {\n 'detect_exact_copies': False,\n 'detect_visual_duplicates': True,\n 'visual_threshold': 10,\n }\n \n try:\n batch_detect_duplicates(self.user, options)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_with_clear_pending(self):\n \"\"\"Test batch detection with clear_pending option.\"\"\"\n # Create existing pending duplicate\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n dup.photos.add(photo1, photo2)\n \n options = {\n 'detect_exact_copies': True,\n 'clear_pending': True,\n }\n \n try:\n batch_detect_duplicates(self.user, options)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_with_null_options(self):\n \"\"\"Test batch detection with None options.\"\"\"\n try:\n batch_detect_duplicates(self.user, None)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_with_empty_options(self):\n \"\"\"Test batch detection with empty options.\"\"\"\n try:\n batch_detect_duplicates(self.user, {})\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n\nclass MultiUserDuplicateIsolationTestCase(TestCase):\n \"\"\"Tests for multi-user duplicate detection isolation.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n\n def test_detection_only_affects_own_photos(self):\n \"\"\"Test that detection only finds duplicates for user's own photos.\"\"\"\n # Create photos for user1 with same image_hash (content hash for deduplication)\n photo1_u1 = create_test_photo(owner=self.user1)\n photo1_u1.image_hash = \"shared_content_hash\"\n photo1_u1.save()\n\n photo2_u1 = create_test_photo(owner=self.user1)\n photo2_u1.image_hash = \"shared_content_hash\"\n photo2_u1.save()\n\n # Create photo for user2 with same image_hash\n photo_u2 = create_test_photo(owner=self.user2)\n photo_u2.image_hash = \"shared_content_hash\"\n photo_u2.save()\n\n # Run detection for user1 only\n detect_exact_copies(self.user1)\n\n # User1 should have duplicates\n _u1_dups = Duplicate.objects.filter(owner=self.user1)\n\n # User2 should have no duplicates\n u2_dups = Duplicate.objects.filter(owner=self.user2)\n self.assertEqual(u2_dups.count(), 0)\n\n def test_clearing_pending_only_affects_own(self):\n \"\"\"Test that clearing pending duplicates only affects user's own.\"\"\"\n # Create duplicates for both users\n for user in [self.user1, self.user2]:\n photo1 = create_test_photo(owner=user)\n photo2 = create_test_photo(owner=user)\n dup = Duplicate.objects.create(\n owner=user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n dup.photos.add(photo1, photo2)\n \n # Run batch detection with clear_pending for user1\n batch_detect_duplicates(self.user1, {'clear_pending': True, 'detect_exact_copies': True})\n \n # User2 should still have their pending duplicate\n u2_pending = Duplicate.objects.filter(\n owner=self.user2,\n review_status=Duplicate.ReviewStatus.PENDING\n )\n self.assertEqual(u2_pending.count(), 1)\n\n\nclass DuplicateCreationEdgeCasesTestCase(TestCase):\n \"\"\"Tests for duplicate creation edge cases.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_three_way_duplicates(self):\n \"\"\"Test handling of three-way duplicates.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n\n # All have same image_hash (content hash for deduplication)\n for photo in [photo1, photo2, photo3]:\n photo.image_hash = \"triple_content_hash\"\n photo.save()\n\n count = detect_exact_copies(self.user)\n\n # Should create one group with 3 photos\n self.assertGreaterEqual(count, 0)\n\n dups = Duplicate.objects.filter(owner=self.user)\n if dups.exists():\n self.assertGreaterEqual(dups.first().photos.count(), 2)\n\n def test_many_duplicates_same_hash(self):\n \"\"\"Test handling of many photos with same hash.\"\"\"\n photos = []\n for i in range(10):\n photo = create_test_photo(owner=self.user)\n photo.image_hash = \"many_duplicates_content_hash\"\n photo.save()\n photos.append(photo)\n\n count = detect_exact_copies(self.user)\n\n # Should create one group with all 10 photos\n self.assertGreaterEqual(count, 0)\n" }, { "path": "api/tests/test_bulk_operations.py", "content": "\"\"\"\nTests for server-side bulk operations (select_all mode).\n\nThese tests verify that bulk operations work correctly when using\nquery-based selection instead of individual image hashes.\n\"\"\"\n\nimport logging\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.models import Photo\nfrom api.tests.utils import create_test_photos, create_test_user\nfrom api.views.photo_filters import build_photo_queryset\n\nlogger = logging.getLogger(__name__)\n\n\nclass BuildPhotoQuerysetTest(TestCase):\n \"\"\"Tests for the build_photo_queryset utility function.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n\n def test_filters_by_owner(self):\n \"\"\"Test that photos are filtered by owner.\"\"\"\n create_test_photos(number_of_photos=3, owner=self.user1)\n create_test_photos(number_of_photos=2, owner=self.user2)\n\n qs = build_photo_queryset(self.user1, {})\n self.assertEqual(qs.count(), 3)\n\n def test_filters_by_video(self):\n \"\"\"Test filtering for videos only.\"\"\"\n create_test_photos(number_of_photos=2, owner=self.user1, video=False)\n create_test_photos(number_of_photos=3, owner=self.user1, video=True)\n\n qs = build_photo_queryset(self.user1, {\"video\": True})\n self.assertEqual(qs.count(), 3)\n\n def test_filters_by_photo(self):\n \"\"\"Test filtering for photos only (non-videos).\"\"\"\n create_test_photos(number_of_photos=2, owner=self.user1, video=False)\n create_test_photos(number_of_photos=3, owner=self.user1, video=True)\n\n qs = build_photo_queryset(self.user1, {\"photo\": True})\n self.assertEqual(qs.count(), 2)\n\n def test_filters_hidden(self):\n \"\"\"Test filtering by hidden status.\"\"\"\n create_test_photos(number_of_photos=2, owner=self.user1, hidden=False)\n create_test_photos(number_of_photos=3, owner=self.user1, hidden=True)\n\n # By default, hidden=False is applied\n qs = build_photo_queryset(self.user1, {})\n self.assertEqual(qs.count(), 2)\n\n # Explicit hidden=True\n qs = build_photo_queryset(self.user1, {\"hidden\": True})\n self.assertEqual(qs.count(), 3)\n\n def test_filters_in_trashcan(self):\n \"\"\"Test filtering by trashcan status.\"\"\"\n create_test_photos(number_of_photos=2, owner=self.user1, in_trashcan=False)\n create_test_photos(number_of_photos=3, owner=self.user1, in_trashcan=True)\n\n # By default, in_trashcan=False is applied\n qs = build_photo_queryset(self.user1, {})\n self.assertEqual(qs.count(), 2)\n\n # Explicit in_trashcan=True\n qs = build_photo_queryset(self.user1, {\"in_trashcan\": True})\n self.assertEqual(qs.count(), 3)\n\n\nclass BulkSetPhotosPublicTest(TestCase):\n \"\"\"Tests for bulk SetPhotosPublic with select_all mode.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_select_all_make_public(self):\n \"\"\"Test making all photos public via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=5, owner=self.user1, public=False)\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"val_public\": True,\n }\n response = self.client.post(\n \"/api/photosedit/makepublic/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 5)\n\n # Verify all photos are now public\n for photo in photos:\n photo.refresh_from_db()\n self.assertTrue(photo.public)\n\n def test_select_all_make_private(self):\n \"\"\"Test making all photos private via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=3, owner=self.user1, public=True)\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"val_public\": False,\n }\n response = self.client.post(\n \"/api/photosedit/makepublic/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3)\n\n # Verify all photos are now private\n for photo in photos:\n photo.refresh_from_db()\n self.assertFalse(photo.public)\n\n def test_select_all_with_exclusions(self):\n \"\"\"Test select_all with some photos excluded.\"\"\"\n photos = create_test_photos(number_of_photos=5, owner=self.user1, public=False)\n excluded_hashes = [photos[0].image_hash, photos[1].image_hash]\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"excluded_hashes\": excluded_hashes,\n \"val_public\": True,\n }\n response = self.client.post(\n \"/api/photosedit/makepublic/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3) # 5 - 2 excluded\n\n # Verify excluded photos are still private\n photos[0].refresh_from_db()\n photos[1].refresh_from_db()\n self.assertFalse(photos[0].public)\n self.assertFalse(photos[1].public)\n\n # Verify other photos are public\n for photo in photos[2:]:\n photo.refresh_from_db()\n self.assertTrue(photo.public)\n\n def test_select_all_only_affects_own_photos(self):\n \"\"\"Test that select_all only affects the user's own photos.\"\"\"\n create_test_photos(number_of_photos=3, owner=self.user1, public=False)\n other_photos = create_test_photos(\n number_of_photos=2, owner=self.user2, public=False\n )\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"val_public\": True,\n }\n response = self.client.post(\n \"/api/photosedit/makepublic/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3)\n\n # Verify other user's photos are untouched\n for photo in other_photos:\n photo.refresh_from_db()\n self.assertFalse(photo.public)\n\n\nclass BulkSetPhotosHiddenTest(TestCase):\n \"\"\"Tests for bulk SetPhotosHidden with select_all mode.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_select_all_hide_photos(self):\n \"\"\"Test hiding all photos via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=4, owner=self.user1, hidden=False)\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"hidden\": True,\n }\n response = self.client.post(\n \"/api/photosedit/hide/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 4)\n\n # Verify all photos are now hidden\n for photo in photos:\n photo.refresh_from_db()\n self.assertTrue(photo.hidden)\n\n def test_select_all_unhide_photos(self):\n \"\"\"Test unhiding all photos via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=3, owner=self.user1, hidden=True)\n\n payload = {\n \"select_all\": True,\n \"query\": {\"hidden\": True}, # Need to query hidden photos\n \"hidden\": False,\n }\n response = self.client.post(\n \"/api/photosedit/hide/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3)\n\n # Verify all photos are now unhidden\n for photo in photos:\n photo.refresh_from_db()\n self.assertFalse(photo.hidden)\n\n\nclass BulkSetPhotosFavoriteTest(TestCase):\n \"\"\"Tests for bulk SetPhotosFavorite with select_all mode.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_select_all_favorite_photos(self):\n \"\"\"Test favoriting all photos via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=4, owner=self.user1, rating=0)\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"favorite\": True,\n }\n response = self.client.post(\n \"/api/photosedit/favorite/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 4)\n\n # Verify all photos are now favorited\n for photo in photos:\n photo.refresh_from_db()\n self.assertGreaterEqual(photo.rating, self.user1.favorite_min_rating)\n\n def test_select_all_unfavorite_photos(self):\n \"\"\"Test unfavoriting all photos via select_all.\"\"\"\n photos = create_test_photos(\n number_of_photos=3,\n owner=self.user1,\n rating=self.user1.favorite_min_rating,\n )\n\n payload = {\n \"select_all\": True,\n \"query\": {\"favorite\": True}, # Need to query favorite photos\n \"favorite\": False,\n }\n response = self.client.post(\n \"/api/photosedit/favorite/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3)\n\n # Verify all photos are now unfavorited\n for photo in photos:\n photo.refresh_from_db()\n self.assertEqual(photo.rating, 0)\n\n\nclass BulkSetPhotosDeletedTest(TestCase):\n \"\"\"Tests for bulk SetPhotosDeleted with select_all mode.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_select_all_move_to_trash(self):\n \"\"\"Test moving all photos to trash via select_all.\"\"\"\n photos = create_test_photos(\n number_of_photos=4, owner=self.user1, in_trashcan=False\n )\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"deleted\": True,\n }\n response = self.client.post(\n \"/api/photosedit/setdeleted/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 4)\n\n # Verify all photos are now in trashcan\n for photo in photos:\n photo.refresh_from_db()\n self.assertTrue(photo.in_trashcan)\n\n def test_select_all_restore_from_trash(self):\n \"\"\"Test restoring all photos from trash via select_all.\"\"\"\n photos = create_test_photos(\n number_of_photos=3, owner=self.user1, in_trashcan=True\n )\n\n payload = {\n \"select_all\": True,\n \"query\": {\"in_trashcan\": True}, # Need to query trashed photos\n \"deleted\": False,\n }\n response = self.client.post(\n \"/api/photosedit/setdeleted/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3)\n\n # Verify all photos are now restored\n for photo in photos:\n photo.refresh_from_db()\n self.assertFalse(photo.in_trashcan)\n\n\nclass BulkSharePhotosTest(TestCase):\n \"\"\"Tests for bulk SetPhotosShared with select_all mode.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_select_all_share_photos(self):\n \"\"\"Test sharing all photos via select_all.\"\"\"\n _photos = create_test_photos(number_of_photos=5, owner=self.user1)\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"val_shared\": True,\n \"target_user_id\": self.user2.id,\n }\n response = self.client.post(\n \"/api/photosedit/share/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 5)\n\n # Verify all photos are shared with user2\n through_model = Photo.shared_to.through\n shared_count = through_model.objects.filter(user_id=self.user2.id).count()\n self.assertEqual(shared_count, 5)\n\n def test_select_all_unshare_photos(self):\n \"\"\"Test unsharing all photos via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=3, owner=self.user1)\n\n # First share the photos (use photo.id, not image_hash, since pk is now UUID)\n through_model = Photo.shared_to.through\n through_model.objects.bulk_create(\n [\n through_model(user_id=self.user2.id, photo_id=p.id)\n for p in photos\n ]\n )\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"val_shared\": False,\n \"target_user_id\": self.user2.id,\n }\n response = self.client.post(\n \"/api/photosedit/share/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 3)\n\n # Verify no photos are shared with user2\n shared_count = through_model.objects.filter(user_id=self.user2.id).count()\n self.assertEqual(shared_count, 0)\n\n def test_select_all_share_with_exclusions(self):\n \"\"\"Test sharing with exclusions via select_all.\"\"\"\n photos = create_test_photos(number_of_photos=5, owner=self.user1)\n excluded_hashes = [photos[0].image_hash]\n\n payload = {\n \"select_all\": True,\n \"query\": {},\n \"excluded_hashes\": excluded_hashes,\n \"val_shared\": True,\n \"target_user_id\": self.user2.id,\n }\n response = self.client.post(\n \"/api/photosedit/share/\", format=\"json\", data=payload\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(data[\"count\"], 4) # 5 - 1 excluded\n\n # Verify excluded photo is not shared (use photo.id, not image_hash)\n through_model = Photo.shared_to.through\n excluded_shared = through_model.objects.filter(\n user_id=self.user2.id, photo_id=photos[0].id\n ).exists()\n self.assertFalse(excluded_shared)\n\n" }, { "path": "api/tests/test_burst_detection_rules.py", "content": "\"\"\"\nComprehensive tests for Burst Detection Rules engine.\n\nTests cover:\n- BurstDetectionRule class functionality\n- Rule condition checking (path, filename, EXIF)\n- Hard criteria rules (EXIF burst mode, sequence number, filename patterns)\n- Soft criteria rules (timestamp proximity, visual similarity)\n- Rule filtering and grouping\n- Edge cases and error handling\n\"\"\"\n\nimport re\nfrom datetime import datetime, timedelta\nfrom unittest.mock import MagicMock, patch\n\nfrom django.test import TestCase\n\nfrom api.burst_detection_rules import (\n BURST_FILENAME_PATTERNS,\n BurstDetectionRule,\n BurstRuleCategory,\n BurstRuleTypes,\n DEFAULT_HARD_RULES,\n DEFAULT_SOFT_RULES,\n as_rules,\n get_all_predefined_burst_rules,\n get_default_burst_detection_rules,\n get_enabled_rules,\n get_hard_rules,\n get_soft_rules,\n group_photos_by_timestamp,\n group_photos_by_visual_similarity,\n)\n\n\nclass BurstRuleTypesTestCase(TestCase):\n \"\"\"Tests for BurstRuleTypes constants.\"\"\"\n\n def test_hard_criteria_types(self):\n \"\"\"Test hard criteria type constants exist.\"\"\"\n self.assertEqual(BurstRuleTypes.EXIF_BURST_MODE, \"exif_burst_mode\")\n self.assertEqual(BurstRuleTypes.EXIF_SEQUENCE_NUMBER, \"exif_sequence_number\")\n self.assertEqual(BurstRuleTypes.FILENAME_PATTERN, \"filename_pattern\")\n\n def test_soft_criteria_types(self):\n \"\"\"Test soft criteria type constants exist.\"\"\"\n self.assertEqual(BurstRuleTypes.TIMESTAMP_PROXIMITY, \"timestamp_proximity\")\n self.assertEqual(BurstRuleTypes.VISUAL_SIMILARITY, \"visual_similarity\")\n\n\nclass BurstRuleCategoryTestCase(TestCase):\n \"\"\"Tests for BurstRuleCategory constants.\"\"\"\n\n def test_categories(self):\n \"\"\"Test category constants.\"\"\"\n self.assertEqual(BurstRuleCategory.HARD, \"hard\")\n self.assertEqual(BurstRuleCategory.SOFT, \"soft\")\n\n\nclass BurstFilenamePatternTestCase(TestCase):\n \"\"\"Tests for predefined filename patterns.\"\"\"\n\n def test_burst_suffix_pattern(self):\n \"\"\"Test _BURST pattern matching.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"burst_suffix\"]\n self.assertIsNotNone(re.search(pattern, \"IMG_001_BURST001\"))\n self.assertIsNotNone(re.search(pattern, \"photo_BURST123\"))\n self.assertIsNone(re.search(pattern, \"IMG_001\"))\n self.assertIsNone(re.search(pattern, \"BURST_photo\"))\n\n def test_sequence_suffix_pattern(self):\n \"\"\"Test sequence number pattern matching.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"sequence_suffix\"]\n self.assertIsNotNone(re.search(pattern, \"IMG_001\"))\n self.assertIsNotNone(re.search(pattern, \"photo_1234\"))\n self.assertIsNone(re.search(pattern, \"IMG_01\")) # Only 2 digits\n self.assertIsNone(re.search(pattern, \"IMG_001_extra\")) # Not at end\n\n def test_bracketed_sequence_pattern(self):\n \"\"\"Test (1), (2) pattern matching.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"bracketed_sequence\"]\n self.assertIsNotNone(re.search(pattern, \"photo (1)\"))\n self.assertIsNotNone(re.search(pattern, \"image (99)\"))\n self.assertIsNotNone(re.search(pattern, \"photo(1)\")) # Pattern allows no space\n self.assertIsNone(re.search(pattern, \"(1) photo\")) # Not at end\n\n def test_samsung_burst_pattern(self):\n \"\"\"Test Samsung burst pattern.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"samsung_burst\"]\n self.assertIsNotNone(re.search(pattern, \"IMG_001_COVER\"))\n self.assertIsNotNone(re.search(pattern, \"photo_123_COVER\"))\n self.assertIsNone(re.search(pattern, \"IMG_COVER\"))\n\n def test_iphone_burst_pattern(self):\n \"\"\"Test iPhone burst pattern.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"iphone_burst\"]\n self.assertIsNotNone(re.search(pattern, \"IMG_1234_5\"))\n self.assertIsNotNone(re.search(pattern, \"IMG_0001_123\"))\n self.assertIsNone(re.search(pattern, \"IMG_123_5\")) # Only 3 digits\n\n\nclass BurstDetectionRuleTestCase(TestCase):\n \"\"\"Tests for BurstDetectionRule class.\"\"\"\n\n def test_create_rule_with_minimal_params(self):\n \"\"\"Test creating rule with minimal required params.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n }\n )\n self.assertEqual(rule.id, 1)\n self.assertEqual(rule.rule_type, BurstRuleTypes.EXIF_BURST_MODE)\n self.assertEqual(rule.name, \"Unnamed rule\")\n self.assertTrue(rule.enabled)\n self.assertTrue(rule.is_default)\n\n def test_create_rule_with_all_params(self):\n \"\"\"Test creating rule with all params.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 42,\n \"name\": \"My Custom Rule\",\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": False,\n \"is_default\": False,\n \"custom_pattern\": r\"_BURST\\d+\",\n }\n )\n self.assertEqual(rule.id, 42)\n self.assertEqual(rule.name, \"My Custom Rule\")\n self.assertEqual(rule.rule_type, BurstRuleTypes.FILENAME_PATTERN)\n self.assertEqual(rule.category, BurstRuleCategory.HARD)\n self.assertFalse(rule.enabled)\n self.assertFalse(rule.is_default)\n self.assertEqual(rule.params[\"custom_pattern\"], r\"_BURST\\d+\")\n\n def test_get_required_exif_tags_burst_mode(self):\n \"\"\"Test required tags for burst mode rule.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n }\n )\n tags = rule.get_required_exif_tags()\n self.assertIn(Tags.BURST_MODE, tags)\n self.assertIn(Tags.CONTINUOUS_DRIVE, tags)\n\n def test_get_required_exif_tags_sequence_number(self):\n \"\"\"Test required tags for sequence number rule.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_SEQUENCE_NUMBER,\n }\n )\n tags = rule.get_required_exif_tags()\n self.assertIn(Tags.SEQUENCE_NUMBER, tags)\n self.assertIn(Tags.IMAGE_NUMBER, tags)\n\n def test_get_required_exif_tags_with_condition(self):\n \"\"\"Test required tags includes condition tag.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"condition_exif\": \"EXIF:Make//Canon\",\n }\n )\n tags = rule.get_required_exif_tags()\n self.assertIn(\"EXIF:Make\", tags)\n\n\nclass RuleConditionTestCase(TestCase):\n \"\"\"Tests for rule condition checking.\"\"\"\n\n def test_check_condition_path_matches(self):\n \"\"\"Test path condition matching.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"condition_path\": r\"/photos/bursts/\",\n }\n )\n self.assertTrue(rule._check_condition_path(\"/photos/bursts/img001.jpg\"))\n self.assertFalse(rule._check_condition_path(\"/photos/normal/img001.jpg\"))\n\n def test_check_condition_path_no_condition(self):\n \"\"\"Test path condition when not set.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n }\n )\n self.assertTrue(rule._check_condition_path(\"/any/path/works.jpg\"))\n\n def test_check_condition_filename_matches(self):\n \"\"\"Test filename condition matching.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"condition_filename\": r\"^IMG_\\d+\",\n }\n )\n self.assertTrue(rule._check_condition_filename(\"/photos/IMG_001.jpg\"))\n self.assertFalse(rule._check_condition_filename(\"/photos/DSC_001.jpg\"))\n\n def test_check_condition_exif_matches(self):\n \"\"\"Test EXIF condition matching.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"condition_exif\": \"EXIF:Make//Canon\",\n }\n )\n self.assertTrue(rule._check_condition_exif({\"EXIF:Make\": \"Canon EOS\"}))\n self.assertFalse(rule._check_condition_exif({\"EXIF:Make\": \"Nikon\"}))\n self.assertFalse(rule._check_condition_exif({}))\n\n def test_check_condition_exif_invalid_format(self):\n \"\"\"Test EXIF condition with invalid format.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"condition_exif\": \"InvalidFormat\", # Missing //\n }\n )\n self.assertFalse(rule._check_condition_exif({\"EXIF:Make\": \"Canon\"}))\n\n def test_check_all_conditions_combined(self):\n \"\"\"Test checking all conditions together.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"condition_path\": r\"/bursts/\",\n \"condition_filename\": r\"^IMG\",\n \"condition_exif\": \"EXIF:Make//Canon\",\n }\n )\n\n # All match\n self.assertTrue(\n rule.check_conditions(\"/bursts/IMG_001.jpg\", {\"EXIF:Make\": \"Canon\"})\n )\n\n # Path doesn't match\n self.assertFalse(\n rule.check_conditions(\"/normal/IMG_001.jpg\", {\"EXIF:Make\": \"Canon\"})\n )\n\n # Filename doesn't match\n self.assertFalse(\n rule.check_conditions(\"/bursts/DSC_001.jpg\", {\"EXIF:Make\": \"Canon\"})\n )\n\n # EXIF doesn't match\n self.assertFalse(\n rule.check_conditions(\"/bursts/IMG_001.jpg\", {\"EXIF:Make\": \"Nikon\"})\n )\n\n\nclass ExifBurstModeRuleTestCase(TestCase):\n \"\"\"Tests for EXIF burst mode detection.\"\"\"\n\n def _create_mock_photo(self, path=\"/photos/test.jpg\", timestamp=None):\n \"\"\"Create a mock photo object.\"\"\"\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = path\n photo.exif_timestamp = timestamp or datetime.now()\n return photo\n\n def test_burst_mode_on(self):\n \"\"\"Test detection with BurstMode = 1.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, group_key = rule.is_burst_photo(photo, {Tags.BURST_MODE: \"1\"})\n\n self.assertTrue(is_burst)\n self.assertIsNotNone(group_key)\n self.assertIn(\"burst_\", group_key)\n\n def test_burst_mode_on_string(self):\n \"\"\"Test detection with BurstMode = 'On'.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {Tags.BURST_MODE: \"On\"})\n self.assertTrue(is_burst)\n\n def test_continuous_drive_on(self):\n \"\"\"Test detection with ContinuousDrive.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {Tags.CONTINUOUS_DRIVE: \"Continuous\"})\n self.assertTrue(is_burst)\n\n def test_burst_mode_off(self):\n \"\"\"Test no detection when BurstMode = 0.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {Tags.BURST_MODE: \"0\"})\n self.assertFalse(is_burst)\n\n def test_disabled_rule_returns_false(self):\n \"\"\"Test disabled rule always returns False.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": False,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {Tags.BURST_MODE: \"1\"})\n self.assertFalse(is_burst)\n\n\nclass ExifSequenceNumberRuleTestCase(TestCase):\n \"\"\"Tests for EXIF sequence number detection.\"\"\"\n\n def _create_mock_photo(self, path=\"/photos/test.jpg\", timestamp=None):\n \"\"\"Create a mock photo object.\"\"\"\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = path\n photo.exif_timestamp = timestamp or datetime.now()\n return photo\n\n def test_sequence_number_detected(self):\n \"\"\"Test detection with valid sequence number.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_SEQUENCE_NUMBER,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, group_key = rule.is_burst_photo(photo, {Tags.SEQUENCE_NUMBER: \"5\"})\n\n self.assertTrue(is_burst)\n self.assertIsNotNone(group_key)\n self.assertIn(\"seq_\", group_key)\n\n def test_sequence_number_zero(self):\n \"\"\"Test detection with sequence number 0.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_SEQUENCE_NUMBER,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {Tags.SEQUENCE_NUMBER: \"0\"})\n self.assertTrue(is_burst)\n\n def test_invalid_sequence_number(self):\n \"\"\"Test no detection with invalid sequence number.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_SEQUENCE_NUMBER,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {Tags.SEQUENCE_NUMBER: \"not_a_number\"})\n self.assertFalse(is_burst)\n\n def test_no_sequence_number(self):\n \"\"\"Test no detection when tag is missing.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_SEQUENCE_NUMBER,\n \"enabled\": True,\n }\n )\n photo = self._create_mock_photo()\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertFalse(is_burst)\n\n\nclass FilenamePatternRuleTestCase(TestCase):\n \"\"\"Tests for filename pattern detection.\"\"\"\n\n def _create_mock_photo(self, path):\n \"\"\"Create a mock photo object.\"\"\"\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = path\n photo.exif_timestamp = datetime.now()\n return photo\n\n def test_burst_suffix_detected(self):\n \"\"\"Test _BURST suffix detection.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n photo = self._create_mock_photo(\"/photos/IMG_001_BURST001.jpg\")\n\n is_burst, group_key = rule.is_burst_photo(photo, {})\n\n self.assertTrue(is_burst)\n self.assertIsNotNone(group_key)\n self.assertIn(\"filename_\", group_key)\n\n def test_sequence_suffix_detected(self):\n \"\"\"Test sequence suffix detection.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n photo = self._create_mock_photo(\"/photos/IMG_001.jpg\")\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertTrue(is_burst)\n\n def test_bracketed_sequence_detected(self):\n \"\"\"Test (1), (2) pattern detection.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n photo = self._create_mock_photo(\"/photos/photo (1).jpg\")\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertTrue(is_burst)\n\n def test_custom_pattern(self):\n \"\"\"Test custom regex pattern.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"custom_pattern\": r\"_HDR\\d+\",\n }\n )\n photo = self._create_mock_photo(\"/photos/IMG_HDR001.jpg\")\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertTrue(is_burst)\n\n def test_specific_pattern_type(self):\n \"\"\"Test specific pattern type selection.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"burst_suffix\",\n }\n )\n\n # Should match burst suffix\n photo1 = self._create_mock_photo(\"/photos/IMG_BURST001.jpg\")\n is_burst, _ = rule.is_burst_photo(photo1, {})\n self.assertTrue(is_burst)\n\n # Should NOT match sequence suffix (different pattern type)\n photo2 = self._create_mock_photo(\"/photos/IMG_001.jpg\")\n is_burst, _ = rule.is_burst_photo(photo2, {})\n self.assertFalse(is_burst)\n\n def test_no_pattern_match(self):\n \"\"\"Test no detection when pattern doesn't match.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"burst_suffix\",\n }\n )\n photo = self._create_mock_photo(\"/photos/normal_photo.jpg\")\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertFalse(is_burst)\n\n def test_no_main_file(self):\n \"\"\"Test handling of photo without main_file.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n }\n )\n photo = MagicMock()\n photo.main_file = None\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertFalse(is_burst)\n\n def test_group_key_contains_directory(self):\n \"\"\"Test group key includes directory for proper grouping.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n\n photo1 = self._create_mock_photo(\"/dir1/IMG_001.jpg\")\n photo2 = self._create_mock_photo(\"/dir2/IMG_001.jpg\")\n\n _, key1 = rule.is_burst_photo(photo1, {})\n _, key2 = rule.is_burst_photo(photo2, {})\n\n # Keys should be different for different directories\n self.assertNotEqual(key1, key2)\n\n\nclass GroupPhotosByTimestampTestCase(TestCase):\n \"\"\"Tests for timestamp proximity grouping.\"\"\"\n\n def _create_mock_photo(self, timestamp, camera_make=\"Canon\", camera_model=\"EOS\"):\n \"\"\"Create a mock photo with timestamp and metadata.\"\"\"\n photo = MagicMock()\n photo.exif_timestamp = timestamp\n photo.metadata = MagicMock()\n photo.metadata.camera_make = camera_make\n photo.metadata.camera_model = camera_model\n return photo\n\n def test_group_consecutive_photos(self):\n \"\"\"Test grouping photos within interval.\"\"\"\n base_time = datetime(2024, 1, 1, 12, 0, 0)\n photos = [\n self._create_mock_photo(base_time),\n self._create_mock_photo(base_time + timedelta(milliseconds=500)),\n self._create_mock_photo(base_time + timedelta(milliseconds=1000)),\n ]\n\n groups = group_photos_by_timestamp(photos, interval_ms=2000)\n\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 3)\n\n def test_separate_groups_by_time_gap(self):\n \"\"\"Test photos with time gap form separate groups.\"\"\"\n base_time = datetime(2024, 1, 1, 12, 0, 0)\n photos = [\n self._create_mock_photo(base_time),\n self._create_mock_photo(base_time + timedelta(milliseconds=500)),\n # Gap of 5 seconds\n self._create_mock_photo(base_time + timedelta(seconds=5)),\n self._create_mock_photo(base_time + timedelta(seconds=5, milliseconds=500)),\n ]\n\n groups = group_photos_by_timestamp(photos, interval_ms=2000)\n\n self.assertEqual(len(groups), 2)\n self.assertEqual(len(groups[0]), 2)\n self.assertEqual(len(groups[1]), 2)\n\n def test_single_photo_not_grouped(self):\n \"\"\"Test single photo doesn't form a group.\"\"\"\n photos = [self._create_mock_photo(datetime.now())]\n\n groups = group_photos_by_timestamp(photos)\n\n self.assertEqual(len(groups), 0)\n\n def test_photos_without_timestamp_skipped(self):\n \"\"\"Test photos without timestamp are skipped.\"\"\"\n base_time = datetime(2024, 1, 1, 12, 0, 0)\n photos = [\n self._create_mock_photo(base_time),\n self._create_mock_photo(None), # No timestamp\n self._create_mock_photo(base_time + timedelta(milliseconds=500)),\n ]\n\n groups = group_photos_by_timestamp(photos)\n\n # Should still group the two with timestamps\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 2)\n\n def test_empty_photo_list(self):\n \"\"\"Test empty input returns empty list.\"\"\"\n groups = group_photos_by_timestamp([])\n self.assertEqual(groups, [])\n\n def test_require_same_camera(self):\n \"\"\"Test same camera requirement.\"\"\"\n base_time = datetime(2024, 1, 1, 12, 0, 0)\n photos = [\n self._create_mock_photo(base_time, \"Canon\", \"EOS\"),\n self._create_mock_photo(\n base_time + timedelta(milliseconds=500), \"Nikon\", \"D850\"\n ),\n self._create_mock_photo(\n base_time + timedelta(milliseconds=1000), \"Canon\", \"EOS\"\n ),\n ]\n\n # With same camera requirement\n groups = group_photos_by_timestamp(photos, require_same_camera=True)\n\n # Different cameras break the group\n self.assertEqual(len(groups), 0)\n\n def test_without_camera_requirement(self):\n \"\"\"Test grouping without camera requirement.\"\"\"\n base_time = datetime(2024, 1, 1, 12, 0, 0)\n photos = [\n self._create_mock_photo(base_time, \"Canon\", \"EOS\"),\n self._create_mock_photo(\n base_time + timedelta(milliseconds=500), \"Nikon\", \"D850\"\n ),\n ]\n\n groups = group_photos_by_timestamp(photos, require_same_camera=False)\n\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 2)\n\n def test_custom_interval(self):\n \"\"\"Test custom interval setting.\"\"\"\n base_time = datetime(2024, 1, 1, 12, 0, 0)\n photos = [\n self._create_mock_photo(base_time),\n self._create_mock_photo(base_time + timedelta(milliseconds=3000)),\n ]\n\n # Default 2000ms interval - should NOT group\n groups = group_photos_by_timestamp(photos, interval_ms=2000)\n self.assertEqual(len(groups), 0)\n\n # 5000ms interval - should group\n groups = group_photos_by_timestamp(photos, interval_ms=5000)\n self.assertEqual(len(groups), 1)\n\n\nclass GroupPhotosByVisualSimilarityTestCase(TestCase):\n \"\"\"Tests for visual similarity grouping.\"\"\"\n\n def _create_mock_photo_with_hash(self, phash):\n \"\"\"Create a mock photo with perceptual hash.\"\"\"\n photo = MagicMock()\n photo.perceptual_hash = phash\n return photo\n\n @patch(\"api.perceptual_hash.hamming_distance\")\n def test_group_similar_photos(self, mock_hamming):\n \"\"\"Test grouping visually similar photos.\"\"\"\n # All photos are similar (distance <= 15)\n mock_hamming.return_value = 5\n\n photos = [\n self._create_mock_photo_with_hash(\"hash1\"),\n self._create_mock_photo_with_hash(\"hash2\"),\n self._create_mock_photo_with_hash(\"hash3\"),\n ]\n\n groups = group_photos_by_visual_similarity(photos, similarity_threshold=15)\n\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 3)\n\n @patch(\"api.perceptual_hash.hamming_distance\")\n def test_separate_dissimilar_photos(self, mock_hamming):\n \"\"\"Test dissimilar photos form separate groups.\"\"\"\n # Distance alternates between similar and dissimilar\n mock_hamming.side_effect = [5, 30, 5] # similar, dissimilar, similar\n\n photos = [\n self._create_mock_photo_with_hash(\"hash1\"),\n self._create_mock_photo_with_hash(\"hash2\"),\n self._create_mock_photo_with_hash(\"hash3\"),\n self._create_mock_photo_with_hash(\"hash4\"),\n ]\n\n groups = group_photos_by_visual_similarity(photos)\n\n # First two group, then third and fourth group separately\n self.assertEqual(len(groups), 2)\n\n def test_photos_without_hash_filtered(self):\n \"\"\"Test photos without hash are filtered out.\"\"\"\n photos = [\n self._create_mock_photo_with_hash(\"hash1\"),\n self._create_mock_photo_with_hash(None),\n self._create_mock_photo_with_hash(\"hash2\"),\n ]\n\n # Mock hamming distance for the remaining two\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=5):\n groups = group_photos_by_visual_similarity(photos)\n\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 2)\n\n def test_empty_list(self):\n \"\"\"Test empty input returns empty list.\"\"\"\n groups = group_photos_by_visual_similarity([])\n self.assertEqual(groups, [])\n\n def test_single_photo_with_hash(self):\n \"\"\"Test single photo doesn't form a group.\"\"\"\n photos = [self._create_mock_photo_with_hash(\"hash1\")]\n\n groups = group_photos_by_visual_similarity(photos)\n self.assertEqual(len(groups), 0)\n\n\nclass DefaultRulesTestCase(TestCase):\n \"\"\"Tests for default rule configurations.\"\"\"\n\n def test_default_hard_rules_count(self):\n \"\"\"Test default hard rules are defined.\"\"\"\n self.assertGreaterEqual(len(DEFAULT_HARD_RULES), 3)\n\n def test_default_soft_rules_count(self):\n \"\"\"Test default soft rules are defined.\"\"\"\n self.assertGreaterEqual(len(DEFAULT_SOFT_RULES), 2)\n\n def test_default_hard_rules_enabled(self):\n \"\"\"Test default hard rules are enabled.\"\"\"\n for rule in DEFAULT_HARD_RULES:\n self.assertTrue(rule.get(\"enabled\", False))\n\n def test_default_soft_rules_disabled(self):\n \"\"\"Test default soft rules are disabled.\"\"\"\n for rule in DEFAULT_SOFT_RULES:\n self.assertFalse(rule.get(\"enabled\", True))\n\n def test_all_default_rules_have_ids(self):\n \"\"\"Test all default rules have unique IDs.\"\"\"\n all_rules = DEFAULT_HARD_RULES + DEFAULT_SOFT_RULES\n ids = [r[\"id\"] for r in all_rules]\n self.assertEqual(len(ids), len(set(ids)))\n\n def test_get_default_burst_detection_rules(self):\n \"\"\"Test getting all default rules.\"\"\"\n rules = get_default_burst_detection_rules()\n self.assertEqual(len(rules), len(DEFAULT_HARD_RULES) + len(DEFAULT_SOFT_RULES))\n\n def test_get_all_predefined_burst_rules(self):\n \"\"\"Test getting all predefined rules including optional.\"\"\"\n all_rules = get_all_predefined_burst_rules()\n self.assertGreater(len(all_rules), len(get_default_burst_detection_rules()))\n\n\nclass RuleFilteringTestCase(TestCase):\n \"\"\"Tests for rule filtering functions.\"\"\"\n\n def test_as_rules(self):\n \"\"\"Test converting configs to rule objects.\"\"\"\n configs = [\n {\"id\": 1, \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE},\n {\"id\": 2, \"rule_type\": BurstRuleTypes.FILENAME_PATTERN},\n ]\n rules = as_rules(configs)\n\n self.assertEqual(len(rules), 2)\n self.assertIsInstance(rules[0], BurstDetectionRule)\n self.assertEqual(rules[0].id, 1)\n\n def test_get_hard_rules(self):\n \"\"\"Test filtering hard rules.\"\"\"\n rules = as_rules(\n [\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": True,\n },\n {\n \"id\": 2,\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": True,\n },\n {\n \"id\": 3,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": False,\n },\n ]\n )\n\n hard_rules = get_hard_rules(rules)\n\n self.assertEqual(len(hard_rules), 1)\n self.assertEqual(hard_rules[0].id, 1)\n\n def test_get_soft_rules(self):\n \"\"\"Test filtering soft rules.\"\"\"\n rules = as_rules(\n [\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": True,\n },\n {\n \"id\": 2,\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": True,\n },\n ]\n )\n\n soft_rules = get_soft_rules(rules)\n\n self.assertEqual(len(soft_rules), 1)\n self.assertEqual(soft_rules[0].id, 2)\n\n def test_get_enabled_rules(self):\n \"\"\"Test filtering enabled rules.\"\"\"\n rules = as_rules(\n [\n {\"id\": 1, \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE, \"enabled\": True},\n {\n \"id\": 2,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": False,\n },\n {\n \"id\": 3,\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"enabled\": True,\n },\n ]\n )\n\n enabled = get_enabled_rules(rules)\n\n self.assertEqual(len(enabled), 2)\n self.assertIn(enabled[0].id, [1, 3])\n self.assertIn(enabled[1].id, [1, 3])\n\n\nclass EdgeCasesTestCase(TestCase):\n \"\"\"Edge case tests for burst detection.\"\"\"\n\n def test_rule_with_none_timestamp(self):\n \"\"\"Test rule handling when photo has no timestamp.\"\"\"\n from api.metadata.tags import Tags\n\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n )\n\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = \"/photos/test.jpg\"\n photo.exif_timestamp = None\n\n is_burst, group_key = rule.is_burst_photo(photo, {Tags.BURST_MODE: \"1\"})\n\n # Should still detect burst, but group_key may be None\n self.assertTrue(is_burst)\n\n def test_group_key_consistency(self):\n \"\"\"Test that same photos always produce same group key.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = \"/photos/IMG_001_BURST001.jpg\"\n photo.exif_timestamp = datetime(2024, 1, 1, 12, 0, 0)\n\n _, key1 = rule.is_burst_photo(photo, {})\n _, key2 = rule.is_burst_photo(photo, {})\n\n self.assertEqual(key1, key2)\n\n def test_empty_exif_tags(self):\n \"\"\"Test handling empty EXIF tags dict.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n )\n\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = \"/photos/test.jpg\"\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertFalse(is_burst)\n\n def test_special_characters_in_filename(self):\n \"\"\"Test filename pattern with special characters.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n\n photo = MagicMock()\n photo.main_file = MagicMock()\n photo.main_file.path = \"/photos/my photo (1).jpg\"\n photo.exif_timestamp = datetime.now()\n\n is_burst, _ = rule.is_burst_photo(photo, {})\n self.assertTrue(is_burst)\n\n def test_case_insensitive_pattern_matching(self):\n \"\"\"Test patterns match case-insensitively.\"\"\"\n rule = BurstDetectionRule(\n {\n \"id\": 1,\n \"rule_type\": BurstRuleTypes.FILENAME_PATTERN,\n \"enabled\": True,\n \"pattern_type\": \"all\",\n }\n )\n\n # Lowercase\n photo1 = MagicMock()\n photo1.main_file = MagicMock()\n photo1.main_file.path = \"/photos/img_burst001.jpg\"\n photo1.exif_timestamp = datetime.now()\n\n # Uppercase\n photo2 = MagicMock()\n photo2.main_file = MagicMock()\n photo2.main_file.path = \"/photos/IMG_BURST001.jpg\"\n photo2.exif_timestamp = datetime.now()\n\n is_burst1, _ = rule.is_burst_photo(photo1, {})\n is_burst2, _ = rule.is_burst_photo(photo2, {})\n\n self.assertTrue(is_burst1)\n self.assertTrue(is_burst2)\n" }, { "path": "api/tests/test_burst_filename_patterns.py", "content": "\"\"\"\nTests for burst detection using filename patterns.\n\nTests:\n- Detection of various filename patterns (_BURST, _001, (1), etc.)\n- Grouping by directory + base name\n- Case insensitivity\n- Edge cases (no extension, special chars, etc.)\n\"\"\"\n\nimport os\nimport re\n\nfrom django.test import TestCase\nfrom django.utils import timezone\n\nfrom api.burst_detection_rules import (\n BURST_FILENAME_PATTERNS,\n check_filename_pattern,\n group_photos_by_timestamp,\n)\nfrom api.models import Photo\nfrom api.models.photo_stack import PhotoStack\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass BurstFilenamePatternMatchingTestCase(TestCase):\n \"\"\"Test filename pattern matching for burst detection.\"\"\"\n\n def test_burst_suffix_pattern(self):\n \"\"\"Test _BURST followed by numbers.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"burst_suffix\"]\n \n # Should match\n self.assertIsNotNone(re.search(pattern, \"IMG_001_BURST001.jpg\", re.IGNORECASE))\n self.assertIsNotNone(re.search(pattern, \"photo_BURST123.jpg\", re.IGNORECASE))\n self.assertIsNotNone(re.search(pattern, \"IMG_BURST99.JPG\", re.IGNORECASE))\n \n # Should not match\n self.assertIsNone(re.search(pattern, \"IMG_001.jpg\", re.IGNORECASE))\n self.assertIsNone(re.search(pattern, \"BURST_photo.jpg\", re.IGNORECASE))\n\n def test_sequence_suffix_pattern(self):\n \"\"\"Test files ending with 3+ digit sequence.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"sequence_suffix\"]\n \n # Should match (need to test on base name without extension)\n base = os.path.splitext(\"IMG_001\")[0]\n self.assertIsNotNone(re.search(pattern, base, re.IGNORECASE))\n \n base = os.path.splitext(\"photo_0001\")[0]\n self.assertIsNotNone(re.search(pattern, base, re.IGNORECASE))\n \n # Should not match\n base = os.path.splitext(\"IMG_01\")[0] # Only 2 digits\n self.assertIsNone(re.search(pattern, base, re.IGNORECASE))\n\n def test_bracketed_sequence_pattern(self):\n \"\"\"Test files with bracketed numbers at end.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"bracketed_sequence\"]\n \n # Should match\n base = os.path.splitext(\"photo (1)\")[0]\n self.assertIsNotNone(re.search(pattern, base, re.IGNORECASE))\n \n base = os.path.splitext(\"IMG (123)\")[0]\n self.assertIsNotNone(re.search(pattern, base, re.IGNORECASE))\n \n # Should not match\n base = os.path.splitext(\"photo [1]\")[0]\n self.assertIsNone(re.search(pattern, base, re.IGNORECASE))\n\n def test_samsung_burst_pattern(self):\n \"\"\"Test Samsung burst cover images.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"samsung_burst\"]\n \n # Should match\n self.assertIsNotNone(re.search(pattern, \"20240101_123456_001_COVER.jpg\", re.IGNORECASE))\n \n # Should not match\n self.assertIsNone(re.search(pattern, \"IMG_001.jpg\", re.IGNORECASE))\n\n def test_iphone_burst_pattern(self):\n \"\"\"Test iPhone burst sequence pattern.\"\"\"\n pattern, _ = BURST_FILENAME_PATTERNS[\"iphone_burst\"]\n \n # Should match\n self.assertIsNotNone(re.search(pattern, \"IMG_1234_1.jpg\", re.IGNORECASE))\n self.assertIsNotNone(re.search(pattern, \"IMG_0001_99.JPG\", re.IGNORECASE))\n \n # Should not match\n self.assertIsNone(re.search(pattern, \"photo_001.jpg\", re.IGNORECASE))\n\n\nclass CheckFilenamePatternTestCase(TestCase):\n \"\"\"Test the check_filename_pattern function.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_check_any_pattern_burst_suffix(self):\n \"\"\"Test detecting burst suffix with any pattern.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001_BURST001.jpg\"\n photo.main_file.save()\n \n matches, group_key = check_filename_pattern(photo, pattern_type=\"any\")\n self.assertTrue(matches)\n self.assertIsNotNone(group_key)\n self.assertIn(\"filename_\", group_key)\n\n def test_check_any_pattern_sequence(self):\n \"\"\"Test detecting sequence suffix with any pattern.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001.jpg\"\n photo.main_file.save()\n \n matches, group_key = check_filename_pattern(photo, pattern_type=\"any\")\n self.assertTrue(matches)\n\n def test_check_any_pattern_bracketed(self):\n \"\"\"Test detecting bracketed sequence with any pattern.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/vacation (1).jpg\"\n photo.main_file.save()\n \n matches, group_key = check_filename_pattern(photo, pattern_type=\"any\")\n self.assertTrue(matches)\n\n def test_check_specific_pattern(self):\n \"\"\"Test checking specific pattern type.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001_BURST001.jpg\"\n photo.main_file.save()\n \n # Should match burst_suffix\n matches, group_key = check_filename_pattern(photo, pattern_type=\"burst_suffix\")\n self.assertTrue(matches)\n \n # Should not match iphone_burst\n matches, group_key = check_filename_pattern(photo, pattern_type=\"iphone_burst\")\n self.assertFalse(matches)\n\n def test_no_match(self):\n \"\"\"Test when no pattern matches.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/random_photo.jpg\"\n photo.main_file.save()\n \n matches, group_key = check_filename_pattern(photo, pattern_type=\"any\")\n self.assertFalse(matches)\n self.assertIsNone(group_key)\n\n def test_group_key_includes_directory(self):\n \"\"\"Test that group key includes directory for grouping.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.main_file.path = \"/photos/2024/IMG_001.jpg\"\n photo1.main_file.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.main_file.path = \"/photos/2023/IMG_001.jpg\"\n photo2.main_file.save()\n \n matches1, key1 = check_filename_pattern(photo1, pattern_type=\"any\")\n matches2, key2 = check_filename_pattern(photo2, pattern_type=\"any\")\n \n self.assertTrue(matches1)\n self.assertTrue(matches2)\n # Different directories = different group keys\n self.assertNotEqual(key1, key2)\n\n def test_same_directory_same_base_grouped(self):\n \"\"\"Test that same directory + base name get same group key.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo1.main_file.path = \"/photos/burst/IMG_001.jpg\"\n photo1.main_file.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.main_file.path = \"/photos/burst/IMG_002.jpg\"\n photo2.main_file.save()\n \n matches1, key1 = check_filename_pattern(photo1, pattern_type=\"any\")\n matches2, key2 = check_filename_pattern(photo2, pattern_type=\"any\")\n \n self.assertTrue(matches1)\n self.assertTrue(matches2)\n # Same directory + same base (IMG) = same group\n # Note: This depends on implementation details\n\n\nclass GroupPhotosByTimestampTestCase(TestCase):\n \"\"\"Test timestamp-based grouping for bursts.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_group_consecutive_timestamps(self):\n \"\"\"Test grouping photos with consecutive timestamps.\"\"\"\n base_time = timezone.now()\n photos = []\n \n for i in range(5):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = base_time + timezone.timedelta(milliseconds=500 * i)\n photo.save()\n photos.append(photo)\n \n # Order by timestamp\n ordered = Photo.objects.filter(pk__in=[p.pk for p in photos]).order_by(\"exif_timestamp\")\n \n groups = group_photos_by_timestamp(ordered, interval_ms=2000)\n \n # All 5 should be in one group (each 500ms apart)\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 5)\n\n def test_separate_groups_by_gap(self):\n \"\"\"Test that large timestamp gaps create separate groups.\"\"\"\n base_time = timezone.now()\n photos = []\n \n # First burst: 3 photos 500ms apart\n for i in range(3):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = base_time + timezone.timedelta(milliseconds=500 * i)\n photo.save()\n photos.append(photo)\n \n # Second burst: 2 photos after 10 second gap\n for i in range(2):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = base_time + timezone.timedelta(seconds=10 + i * 0.5)\n photo.save()\n photos.append(photo)\n \n ordered = Photo.objects.filter(pk__in=[p.pk for p in photos]).order_by(\"exif_timestamp\")\n \n groups = group_photos_by_timestamp(ordered, interval_ms=2000)\n \n # Should be 2 groups\n self.assertEqual(len(groups), 2)\n self.assertEqual(len(groups[0]), 3)\n self.assertEqual(len(groups[1]), 2)\n\n def test_single_photo_no_group(self):\n \"\"\"Test that single photos don't form groups.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = timezone.now()\n photo.save()\n \n ordered = Photo.objects.filter(pk=photo.pk).order_by(\"exif_timestamp\")\n \n groups = group_photos_by_timestamp(ordered, interval_ms=2000)\n \n # Single photo should not form a group\n self.assertEqual(len(groups), 0)\n\n def test_empty_queryset(self):\n \"\"\"Test with empty queryset.\"\"\"\n ordered = Photo.objects.none()\n \n groups = group_photos_by_timestamp(ordered, interval_ms=2000)\n \n self.assertEqual(len(groups), 0)\n\n def test_photos_without_timestamp(self):\n \"\"\"Test handling photos without exif_timestamp.\"\"\"\n photos = []\n for _ in range(3):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = None\n photo.save()\n photos.append(photo)\n \n ordered = Photo.objects.filter(pk__in=[p.pk for p in photos]).order_by(\"exif_timestamp\")\n \n groups = group_photos_by_timestamp(ordered, interval_ms=2000)\n \n # Photos without timestamps can't be grouped by timestamp\n self.assertEqual(len(groups), 0)\n\n\nclass BurstDetectionIntegrationTestCase(TestCase):\n \"\"\"Integration tests for burst detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_detect_burst_creates_stack(self):\n \"\"\"Test that detecting a burst creates a stack.\"\"\"\n from api.stack_detection import detect_burst_sequences\n \n base_time = timezone.now()\n photos = []\n \n for i in range(4):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = base_time + timezone.timedelta(milliseconds=300 * i)\n photo.main_file.path = f\"/photos/burst/IMG_{i:03d}.jpg\"\n photo.main_file.save()\n photo.save()\n photos.append(photo)\n \n # Run burst detection\n detect_burst_sequences(self.user)\n \n # Check for burst stacks\n _burst_stacks = PhotoStack.objects.filter(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n )\n \n # Should have created at least one burst stack\n # (depends on detection rules being enabled)\n\n def test_case_insensitive_pattern_matching(self):\n \"\"\"Test that filename patterns are case-insensitive.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001_BURST001.JPG\" # Uppercase extension\n photo.main_file.save()\n \n matches, _ = check_filename_pattern(photo, pattern_type=\"any\")\n self.assertTrue(matches)\n \n photo2 = create_test_photo(owner=self.user)\n photo2.main_file.path = \"/photos/img_001_burst001.jpg\" # Lowercase\n photo2.main_file.save()\n \n matches2, _ = check_filename_pattern(photo2, pattern_type=\"any\")\n self.assertTrue(matches2)\n\n\nclass FilenamePatternEdgeCasesTestCase(TestCase):\n \"\"\"Test edge cases for filename pattern detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_no_extension(self):\n \"\"\"Test handling files without extension.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001\" # No extension\n photo.main_file.save()\n \n matches, _ = check_filename_pattern(photo, pattern_type=\"any\")\n # Should still match based on base name\n self.assertTrue(matches)\n\n def test_multiple_extensions(self):\n \"\"\"Test handling files with multiple dots.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001.edited.jpg\"\n photo.main_file.save()\n \n matches, _ = check_filename_pattern(photo, pattern_type=\"any\")\n # May or may not match depending on extension handling\n\n def test_unicode_filename(self):\n \"\"\"Test handling unicode filenames.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/写真_001.jpg\"\n photo.main_file.save()\n \n matches, _ = check_filename_pattern(photo, pattern_type=\"any\")\n # Should handle gracefully\n\n def test_very_long_filename(self):\n \"\"\"Test handling very long filenames.\"\"\"\n photo = create_test_photo(owner=self.user)\n long_name = \"A\" * 200 + \"_001.jpg\"\n photo.main_file.path = f\"/photos/{long_name}\"\n photo.main_file.save()\n \n matches, _ = check_filename_pattern(photo, pattern_type=\"any\")\n # Should handle gracefully\n\n def test_special_characters_in_path(self):\n \"\"\"Test handling special characters in path.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/my folder (2024)/IMG_001.jpg\"\n photo.main_file.save()\n \n matches, _ = check_filename_pattern(photo, pattern_type=\"any\")\n self.assertTrue(matches)\n\n def test_invalid_pattern_type(self):\n \"\"\"Test handling invalid pattern type.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file.path = \"/photos/IMG_001_BURST001.jpg\"\n photo.main_file.save()\n \n # Invalid pattern type should not match\n matches, _ = check_filename_pattern(photo, pattern_type=\"nonexistent_pattern\")\n self.assertFalse(matches)\n" }, { "path": "api/tests/test_delete_photos.py", "content": "from unittest.mock import patch\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import create_test_photos, create_test_user\n\n\nclass DeletePhotosTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_tag_my_photos_for_removal(self):\n photos = create_test_photos(number_of_photos=3, owner=self.user1)\n image_hashes = [p.image_hash for p in photos]\n\n payload = {\"image_hashes\": image_hashes, \"deleted\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/setdeleted/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(3, len(data[\"results\"]))\n self.assertEqual(3, len(data[\"updated\"]))\n self.assertEqual(0, len(data[\"not_updated\"]))\n\n def test_untag_my_photos_for_removal(self):\n photos1 = create_test_photos(\n number_of_photos=1, owner=self.user1, in_trashcan=True\n )\n photos2 = create_test_photos(number_of_photos=2, owner=self.user1)\n image_hashes = [p.image_hash for p in photos1 + photos2]\n\n payload = {\"image_hashes\": image_hashes, \"deleted\": False}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/setdeleted/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(1, len(data[\"results\"]))\n self.assertEqual(1, len(data[\"updated\"]))\n self.assertEqual(2, len(data[\"not_updated\"]))\n\n def test_tag_photos_of_other_user_for_removal(self):\n photos = create_test_photos(number_of_photos=2, owner=self.user2)\n image_hashes = [p.image_hash for p in photos]\n\n payload = {\"image_hashes\": image_hashes, \"deleted\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/setdeleted/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"updated\"]))\n # Photos not owned by user are treated as \"missing\" for security (no info leak)\n self.assertEqual(0, len(data[\"not_updated\"]))\n\n @patch(\"api.views.photos.logger.warning\", autospec=True)\n def test_tag_for_removal_nonexistent_photo(self, logger):\n payload = {\"image_hashes\": [\"nonexistent_photo\"], \"deleted\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/setdeleted/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"updated\"]))\n self.assertEqual(0, len(data[\"not_updated\"]))\n logger.assert_called_with(\n \"Could not set photo nonexistent_photo to deleted. It does not exist or is not owned by user.\"\n )\n\n def test_delete_tagged_photos_for_removal(self):\n photos_to_delete = create_test_photos(\n number_of_photos=2, owner=self.user1, in_trashcan=True\n )\n photos_to_not_delete = create_test_photos(number_of_photos=3, owner=self.user1)\n image_hashes = [p.image_hash for p in photos_to_delete + photos_to_not_delete]\n\n payload = {\"image_hashes\": image_hashes}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.delete(\n \"/api/photosedit/delete/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(2, len(data[\"results\"]))\n self.assertEqual(2, len(data[\"deleted\"]))\n self.assertEqual(3, len(data[\"not_deleted\"]))\n\n def test_delete_tagged_photos_of_other_user_for_removal(self):\n photos_to_delete = create_test_photos(\n number_of_photos=5, owner=self.user2, in_trashcan=True\n )\n image_hashes = [p.image_hash for p in photos_to_delete]\n\n payload = {\"image_hashes\": image_hashes}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.delete(\n \"/api/photosedit/delete/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"deleted\"]))\n self.assertEqual(5, len(data[\"not_deleted\"]))\n" }, { "path": "api/tests/test_detection_edge_cases.py", "content": "\"\"\"\nEdge case tests for detection logic.\n\nTests cover:\n- Burst detection rule parsing and validation\n- File path and naming edge cases\n- Detection with missing/corrupt data\n\nNOTE: RAW+JPEG and Live Photo detection tests are skipped because these\nfunctions were removed. RAW+JPEG and Live Photos now use the file variants\nmodel (Photo.files) instead of stacks, handled during scan time.\n\"\"\"\n\nimport json\nimport unittest\nfrom datetime import datetime, timedelta\nfrom django.test import TestCase\n\nfrom api.models.file import File\nfrom api.models.photo_stack import PhotoStack\nfrom api.burst_detection_rules import (\n BurstDetectionRule,\n BurstRuleTypes,\n BurstRuleCategory,\n BURST_FILENAME_PATTERNS,\n get_default_burst_detection_rules,\n as_rules,\n)\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass BurstRuleParsingTestCase(TestCase):\n \"\"\"Tests for burst rule parsing and validation.\"\"\"\n\n def test_parse_valid_rule(self):\n \"\"\"Test parsing a valid burst rule.\"\"\"\n rule_params = {\n \"id\": \"test_rule\",\n \"name\": \"Test Rule\",\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"category\": BurstRuleCategory.HARD,\n \"enabled\": True,\n }\n rule = BurstDetectionRule(rule_params)\n \n self.assertEqual(rule.id, \"test_rule\")\n self.assertEqual(rule.name, \"Test Rule\")\n self.assertEqual(rule.rule_type, BurstRuleTypes.EXIF_BURST_MODE)\n self.assertTrue(rule.enabled)\n\n def test_parse_rule_with_missing_optional_fields(self):\n \"\"\"Test parsing rule with only required fields.\"\"\"\n rule_params = {\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n }\n rule = BurstDetectionRule(rule_params)\n \n self.assertIsNone(rule.id)\n self.assertEqual(rule.name, \"Unnamed rule\")\n self.assertEqual(rule.category, BurstRuleCategory.HARD)\n self.assertTrue(rule.enabled) # Default True\n self.assertTrue(rule.is_default) # Default True\n\n def test_parse_disabled_rule(self):\n \"\"\"Test parsing a disabled rule.\"\"\"\n rule_params = {\n \"id\": \"disabled_rule\",\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"enabled\": False,\n }\n rule = BurstDetectionRule(rule_params)\n \n self.assertFalse(rule.enabled)\n\n def test_default_rules_are_valid(self):\n \"\"\"Test that all default rules parse correctly.\"\"\"\n default_rules = get_default_burst_detection_rules()\n for rule_dict in default_rules:\n rule = BurstDetectionRule(rule_dict)\n self.assertIsNotNone(rule.rule_type)\n self.assertIn(rule.category, [BurstRuleCategory.HARD, BurstRuleCategory.SOFT])\n\n\nclass BurstFilenamePatternTestCase(TestCase):\n \"\"\"Tests for burst filename pattern matching.\"\"\"\n\n def test_burst_suffix_pattern(self):\n \"\"\"Test _BURST pattern matching.\"\"\"\n import re\n pattern = BURST_FILENAME_PATTERNS[\"burst_suffix\"][0]\n \n # Should match\n self.assertIsNotNone(re.search(pattern, \"IMG_001_BURST001\"))\n self.assertIsNotNone(re.search(pattern, \"photo_BURST123\"))\n \n # Should not match\n self.assertIsNone(re.search(pattern, \"IMG_001\"))\n self.assertIsNone(re.search(pattern, \"BURST_photo\"))\n\n def test_sequence_suffix_pattern(self):\n \"\"\"Test sequence number suffix pattern.\"\"\"\n import re\n pattern = BURST_FILENAME_PATTERNS[\"sequence_suffix\"][0]\n \n # Should match\n self.assertIsNotNone(re.search(pattern, \"IMG_001\"))\n self.assertIsNotNone(re.search(pattern, \"photo_0001\"))\n \n # Should not match (less than 3 digits)\n self.assertIsNone(re.search(pattern, \"IMG_01\"))\n\n def test_bracketed_sequence_pattern(self):\n \"\"\"Test bracketed sequence pattern.\"\"\"\n import re\n pattern = BURST_FILENAME_PATTERNS[\"bracketed_sequence\"][0]\n \n # Should match\n self.assertIsNotNone(re.search(pattern, \"photo (1)\"))\n self.assertIsNotNone(re.search(pattern, \"image (123)\"))\n \n # Should not match\n self.assertIsNone(re.search(pattern, \"photo\"))\n self.assertIsNone(re.search(pattern, \"(1) photo\"))\n\n\nclass UserBurstRulesTestCase(TestCase):\n \"\"\"Tests for user burst detection rules configuration.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_as_rules_with_default_rules(self):\n \"\"\"Test as_rules with default rules config.\"\"\"\n default_config = get_default_burst_detection_rules()\n rules = as_rules(default_config)\n \n self.assertIsInstance(rules, list)\n self.assertGreater(len(rules), 0)\n\n def test_as_rules_with_custom_rules(self):\n \"\"\"Test as_rules with custom rule config.\"\"\"\n custom_rules = [\n {\n \"id\": \"custom1\",\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n ]\n \n rules = as_rules(custom_rules)\n \n self.assertEqual(len(rules), 1)\n self.assertEqual(rules[0].id, \"custom1\")\n\n def test_as_rules_with_empty_list(self):\n \"\"\"Test as_rules with empty list.\"\"\"\n rules = as_rules([])\n \n self.assertEqual(len(rules), 0)\n\n def test_user_rules_stored_as_json(self):\n \"\"\"Test that user rules can be stored as JSON string.\"\"\"\n custom_rules = [\n {\n \"id\": \"custom1\",\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": True,\n }\n ]\n self.user.burst_detection_rules = json.dumps(custom_rules)\n self.user.save()\n \n # Reload and parse\n self.user.refresh_from_db()\n rules_config = json.loads(self.user.burst_detection_rules)\n rules = as_rules(rules_config)\n \n self.assertEqual(len(rules), 1)\n\n\n@unittest.skip(\"RAW+JPEG detection removed - now handled via file variants during scan\")\nclass RawJpegDetectionEdgeCasesTestCase(TestCase):\n \"\"\"Tests for RAW+JPEG detection edge cases.\n \n DEPRECATED: RAW+JPEG pairs are now handled as file variants (Photo.files)\n during scan time, not via stack detection. These tests are skipped.\n \"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_detection_with_no_raw_photos(self):\n \"\"\"Test RAW+JPEG detection when there are no RAW files.\"\"\"\n from api.stack_detection import detect_raw_jpeg_pairs\n \n # Create only JPEG photos (type=1 is IMAGE)\n for i in range(3):\n photo = create_test_photo(owner=self.user)\n photo.main_file.type = File.IMAGE\n photo.main_file.save()\n \n stacks_created = detect_raw_jpeg_pairs(self.user)\n \n self.assertEqual(stacks_created, 0)\n\n def test_detection_with_raw_no_matching_jpeg(self):\n \"\"\"Test RAW+JPEG detection when RAW has no matching JPEG.\"\"\"\n from api.stack_detection import detect_raw_jpeg_pairs\n \n # Create RAW photo\n raw_photo = create_test_photo(owner=self.user)\n raw_photo.main_file.type = File.RAW_FILE\n raw_photo.main_file.path = \"/photos/unique_raw.CR2\"\n raw_photo.main_file.save()\n \n # Create JPEG with different name\n jpeg_photo = create_test_photo(owner=self.user)\n jpeg_photo.main_file.type = File.IMAGE\n jpeg_photo.main_file.path = \"/photos/different_name.jpg\"\n jpeg_photo.main_file.save()\n \n stacks_created = detect_raw_jpeg_pairs(self.user)\n \n self.assertEqual(stacks_created, 0)\n\n def test_detection_case_insensitive_extensions(self):\n \"\"\"Test that RAW+JPEG detection handles case variations.\"\"\"\n from api.stack_detection import detect_raw_jpeg_pairs\n \n # Create RAW photo\n raw_photo = create_test_photo(owner=self.user)\n raw_photo.main_file.type = File.RAW_FILE\n raw_photo.main_file.path = \"/photos/image.CR2\"\n raw_photo.main_file.save()\n \n # Create JPEG with uppercase extension\n jpeg_photo = create_test_photo(owner=self.user)\n jpeg_photo.main_file.type = File.IMAGE\n jpeg_photo.main_file.path = \"/photos/image.JPG\"\n jpeg_photo.main_file.save()\n \n stacks_created = detect_raw_jpeg_pairs(self.user)\n \n # Should find the pair regardless of case\n self.assertGreaterEqual(stacks_created, 0)\n\n def test_detection_with_photo_no_main_file(self):\n \"\"\"Test detection handles photos without main_file.\"\"\"\n from api.stack_detection import detect_raw_jpeg_pairs\n \n # Create a regular test photo (which has main_file)\n # Then manually set main_file to None to simulate edge case\n photo = create_test_photo(owner=self.user)\n photo.main_file = None\n photo.save()\n \n # Should not crash\n stacks_created = detect_raw_jpeg_pairs(self.user)\n self.assertGreaterEqual(stacks_created, 0)\n\n def test_detection_clears_existing_stacks(self):\n \"\"\"Test that re-detection clears existing RAW+JPEG stacks.\"\"\"\n from api.stack_detection import detect_raw_jpeg_pairs\n \n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n # Create existing RAW+JPEG stack\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.RAW_JPEG_PAIR,\n )\n stack.photos.add(photo1, photo2)\n \n initial_count = PhotoStack.objects.filter(\n owner=self.user,\n stack_type=PhotoStack.StackType.RAW_JPEG_PAIR\n ).count()\n self.assertEqual(initial_count, 1)\n \n # Run detection\n detect_raw_jpeg_pairs(self.user)\n \n # Old stack should be cleared\n # (new stacks may or may not be created depending on file types)\n\n\nclass BurstDetectionEdgeCasesTestCase(TestCase):\n \"\"\"Tests for burst detection edge cases.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_detection_with_no_photos(self):\n \"\"\"Test burst detection with empty library.\"\"\"\n from api.stack_detection import detect_burst_sequences\n \n stacks_created = detect_burst_sequences(self.user)\n \n self.assertEqual(stacks_created, 0)\n\n def test_detection_with_all_rules_disabled(self):\n \"\"\"Test burst detection when all rules are disabled.\"\"\"\n from api.stack_detection import detect_burst_sequences\n \n # Create some photos\n for i in range(3):\n create_test_photo(owner=self.user)\n \n # Disable all rules\n disabled_rules = [\n {\n \"id\": \"disabled1\",\n \"rule_type\": BurstRuleTypes.EXIF_BURST_MODE,\n \"enabled\": False,\n }\n ]\n self.user.burst_detection_rules = json.dumps(disabled_rules)\n self.user.save()\n \n stacks_created = detect_burst_sequences(self.user)\n \n # No stacks should be created with all rules disabled\n self.assertEqual(stacks_created, 0)\n\n def test_detection_with_trashed_photos_excluded(self):\n \"\"\"Test that trashed photos are excluded from detection.\"\"\"\n from api.stack_detection import detect_burst_sequences\n \n # Create photos, some in trash\n _photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo2.in_trashcan = True\n photo2.save()\n \n _stacks_created = detect_burst_sequences(self.user)\n \n # Trashed photos should not be in any stack\n stacks = PhotoStack.objects.filter(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n )\n for stack in stacks:\n self.assertFalse(stack.photos.filter(in_trashcan=True).exists())\n\n\nclass TimestampProximityRuleTestCase(TestCase):\n \"\"\"Tests for timestamp proximity burst detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_photos_within_threshold_grouped(self):\n \"\"\"Test that photos within timestamp threshold are grouped.\"\"\"\n from django.utils import timezone\n from api.stack_detection import _detect_bursts_soft_criteria\n \n base_time = timezone.make_aware(datetime(2024, 1, 1, 12, 0, 0))\n \n # Create photos with close timestamps\n photo1 = create_test_photo(owner=self.user)\n photo1.exif_timestamp = base_time\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.exif_timestamp = base_time + timedelta(seconds=1)\n photo2.save()\n \n photo3 = create_test_photo(owner=self.user)\n photo3.exif_timestamp = base_time + timedelta(seconds=2)\n photo3.save()\n \n # Soft criteria with 3000ms interval\n soft_rules = [\n BurstDetectionRule({\n \"id\": \"timestamp\",\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": True,\n \"interval_ms\": 3000,\n })\n ]\n \n stacks_created = _detect_bursts_soft_criteria(self.user, soft_rules)\n \n # Should group the 3 photos\n self.assertGreaterEqual(stacks_created, 0)\n\n def test_photos_beyond_threshold_not_grouped(self):\n \"\"\"Test that photos beyond timestamp threshold are not grouped.\"\"\"\n from django.utils import timezone\n from api.stack_detection import _detect_bursts_soft_criteria\n \n base_time = timezone.make_aware(datetime(2024, 1, 1, 12, 0, 0))\n \n # Create photos with far timestamps\n photo1 = create_test_photo(owner=self.user)\n photo1.exif_timestamp = base_time\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.exif_timestamp = base_time + timedelta(minutes=5)\n photo2.save()\n \n # Soft criteria with 3000ms interval\n soft_rules = [\n BurstDetectionRule({\n \"id\": \"timestamp\",\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": True,\n \"interval_ms\": 3000,\n })\n ]\n \n stacks_created = _detect_bursts_soft_criteria(self.user, soft_rules)\n \n # Should not group photos that are 5 minutes apart\n self.assertEqual(stacks_created, 0)\n\n def test_photos_without_timestamp_skipped(self):\n \"\"\"Test that photos without timestamp are skipped.\"\"\"\n from api.stack_detection import _detect_bursts_soft_criteria\n \n # Create photo without timestamp\n photo1 = create_test_photo(owner=self.user)\n photo1.exif_timestamp = None\n photo1.save()\n \n photo2 = create_test_photo(owner=self.user)\n photo2.exif_timestamp = None\n photo2.save()\n \n soft_rules = [\n BurstDetectionRule({\n \"id\": \"timestamp\",\n \"rule_type\": BurstRuleTypes.TIMESTAMP_PROXIMITY,\n \"category\": BurstRuleCategory.SOFT,\n \"enabled\": True,\n \"interval_ms\": 3000,\n })\n ]\n \n # Should not crash - function filters photos without timestamps\n stacks_created = _detect_bursts_soft_criteria(self.user, soft_rules)\n self.assertEqual(stacks_created, 0)\n\n\n@unittest.skip(\"Live Photo detection removed - now handled via file variants during scan\")\nclass LivePhotoDetectionEdgeCasesTestCase(TestCase):\n \"\"\"Tests for live photo detection edge cases.\n \n DEPRECATED: Live Photos are now handled as file variants (Photo.files)\n during scan time, not via stack detection. These tests are skipped.\n \"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_detection_with_no_live_photos(self):\n \"\"\"Test live photo detection with no live photos.\"\"\"\n from api.stack_detection import detect_live_photos\n \n # Create regular photos\n for i in range(3):\n create_test_photo(owner=self.user)\n \n stacks_created = detect_live_photos(self.user)\n \n self.assertEqual(stacks_created, 0)\n\n\nclass DetectionProgressCallbackTestCase(TestCase):\n \"\"\"Tests for detection progress callbacks.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n @unittest.skip(\"RAW+JPEG detection removed - now handled via file variants during scan\")\n def test_raw_jpeg_detection_calls_progress(self):\n \"\"\"Test that RAW+JPEG detection calls progress callback.\n \n DEPRECATED: RAW+JPEG pairs are now handled as file variants (Photo.files)\n during scan time, not via stack detection.\n \"\"\"\n from api.stack_detection import detect_raw_jpeg_pairs\n \n # Create some RAW photos\n for i in range(3):\n photo = create_test_photo(owner=self.user)\n photo.main_file.type = File.RAW_FILE\n photo.main_file.save()\n \n progress_calls = []\n \n def progress_callback(current, total, found):\n progress_calls.append((current, total, found))\n \n detect_raw_jpeg_pairs(self.user, progress_callback=progress_callback)\n \n # Progress should have been called\n self.assertGreater(len(progress_calls), 0)\n\n def test_burst_detection_calls_progress(self):\n \"\"\"Test that burst detection calls progress callback.\"\"\"\n from api.stack_detection import detect_burst_sequences\n \n # Create some photos\n for i in range(3):\n create_test_photo(owner=self.user)\n \n progress_calls = []\n \n def progress_callback(current, total, found):\n progress_calls.append((current, total, found))\n \n detect_burst_sequences(self.user, progress_callback=progress_callback)\n \n # Progress may or may not be called depending on implementation\n # Just verify it doesn't crash\n self.assertIsInstance(progress_calls, list)\n\n\nclass BatchDetectionEdgeCasesTestCase(TestCase):\n \"\"\"Tests for batch detection function.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_batch_detection_all_types(self):\n \"\"\"Test batch detection with all detection types enabled.\"\"\"\n from api.stack_detection import batch_detect_stacks\n \n options = {\n 'detect_raw_jpeg': True,\n 'detect_bursts': True,\n 'detect_live_photos': True,\n }\n \n # Should not crash - function may return None (runs as job)\n try:\n batch_detect_stacks(self.user, options)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_none_enabled(self):\n \"\"\"Test batch detection with no detection types enabled.\"\"\"\n from api.stack_detection import batch_detect_stacks\n \n options = {\n 'detect_raw_jpeg': False,\n 'detect_bursts': False,\n 'detect_live_photos': False,\n }\n \n # Should not crash\n try:\n batch_detect_stacks(self.user, options)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_with_null_options(self):\n \"\"\"Test batch detection with None options.\"\"\"\n from api.stack_detection import batch_detect_stacks\n \n # Should use defaults and not crash\n try:\n batch_detect_stacks(self.user, None)\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n def test_batch_detection_with_empty_options(self):\n \"\"\"Test batch detection with empty options dict.\"\"\"\n from api.stack_detection import batch_detect_stacks\n \n # Should not crash\n try:\n batch_detect_stacks(self.user, {})\n success = True\n except Exception:\n success = False\n \n self.assertTrue(success)\n\n\nclass MultiUserDetectionIsolationTestCase(TestCase):\n \"\"\"Tests for multi-user detection isolation.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n\n def test_detection_only_affects_own_photos(self):\n \"\"\"Test that detection only creates stacks for user's own photos.\"\"\"\n from api.stack_detection import detect_burst_sequences\n \n # Create photos for both users\n for i in range(3):\n create_test_photo(owner=self.user1)\n create_test_photo(owner=self.user2)\n \n # Run detection for user1 only\n detect_burst_sequences(self.user1)\n \n # User2's photos should not have any stacks\n user2_stacks = PhotoStack.objects.filter(owner=self.user2)\n self.assertEqual(user2_stacks.count(), 0)\n\n def test_clearing_stacks_only_affects_own(self):\n \"\"\"Test that clearing stacks only affects user's own stacks.\"\"\"\n from api.stack_detection import clear_stacks_of_type\n \n # Create stacks for both users\n for user in [self.user1, self.user2]:\n photo1 = create_test_photo(owner=user)\n photo2 = create_test_photo(owner=user)\n stack = PhotoStack.objects.create(\n owner=user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(photo1, photo2)\n \n # Clear stacks for user1 only\n clear_stacks_of_type(self.user1, PhotoStack.StackType.BURST_SEQUENCE)\n \n # User1 should have no stacks\n user1_stacks = PhotoStack.objects.filter(\n owner=self.user1,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n )\n self.assertEqual(user1_stacks.count(), 0)\n \n # User2 should still have their stack\n user2_stacks = PhotoStack.objects.filter(\n owner=self.user2,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE\n )\n self.assertEqual(user2_stacks.count(), 1)\n" }, { "path": "api/tests/test_directory_watcher_fix.py", "content": "from django.test import TestCase\nfrom django.db.models import Q\n\nfrom api.models import Photo\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass DirectoryWatcherFixTest(TestCase):\n def setUp(self):\n self.user = create_test_user()\n\n def test_generate_tags_query_works(self):\n \"\"\"Test that the generate_tags query works with the new PhotoCaption model\"\"\"\n # Create a photo without places365 captions\n photo = create_test_photo(owner=self.user)\n\n # Add some caption data to the photo (but NOT places365)\n from api.models.photo_caption import PhotoCaption\n\n caption_instance, created = PhotoCaption.objects.get_or_create(photo=photo)\n caption_instance.captions_json = {\n \"im2txt\": \"A beautiful landscape\",\n \"user_caption\": \"My vacation photo\",\n }\n caption_instance.save()\n\n # This query should work without FieldError\n existing_photos = Photo.objects.filter(\n Q(owner=self.user.id)\n & (\n Q(caption_instance__isnull=True)\n | Q(caption_instance__captions_json__isnull=True)\n | Q(caption_instance__captions_json__places365__isnull=True)\n )\n )\n\n # Should find the photo since it has no places365 captions\n self.assertEqual(existing_photos.count(), 1)\n self.assertEqual(existing_photos.first(), photo)\n\n def test_generate_tags_query_excludes_photos_with_places365(self):\n \"\"\"Test that photos with places365 captions are excluded\"\"\"\n # Create a photo with places365 captions\n photo = create_test_photo(owner=self.user)\n from api.models.photo_caption import PhotoCaption\n\n caption_instance, created = PhotoCaption.objects.get_or_create(photo=photo)\n caption_instance.captions_json = {\n \"places365\": {\"categories\": [\"outdoor\"], \"attributes\": [\"sunny\"]}\n }\n caption_instance.save()\n\n # This query should exclude the photo since it has places365 captions\n existing_photos = Photo.objects.filter(\n Q(owner=self.user.id)\n & (\n Q(caption_instance__isnull=True)\n | Q(caption_instance__captions_json__isnull=True)\n | Q(caption_instance__captions_json__places365__isnull=True)\n )\n )\n\n # Should not find the photo since it has places365 captions\n self.assertEqual(existing_photos.count(), 0)\n" }, { "path": "api/tests/test_dirtree.py", "content": "from django.test import TestCase\nfrom pyfakefs.fake_filesystem_unittest import Patcher\nfrom rest_framework.test import APIClient\n\nfrom api.models import User\nfrom api.tests.utils import create_password\n\n\nclass DirTreeTest(TestCase):\n def setUp(self):\n self.admin = User.objects.create_superuser(\n \"admin\", \"admin@test.com\", create_password()\n )\n self.user = User.objects.create_user(\"user\", \"user@test.com\", create_password())\n self.client = APIClient()\n\n def test_admin_should_allow_to_retrieve_dirtree(self):\n self.client.force_authenticate(user=self.admin)\n response = self.client.get(\"/api/dirtree/\")\n self.assertEqual(200, response.status_code)\n\n def test_should_retrieve_dir_listing_by_path(self):\n self.client.force_authenticate(user=self.admin)\n response = self.client.get(\"/api/dirtree/?path=/data\")\n self.assertEqual(200, response.status_code)\n\n def test_should_fail_when_listing_with_invalid_path(self):\n self.client.force_authenticate(user=self.admin)\n response = self.client.get(\"/api/dirtree/?path=/does_not_exist\")\n data = response.json()\n self.assertEqual(403, response.status_code)\n self.assertEqual(\n data[\"message\"], \"Access denied. Path is outside the allowed directory.\"\n )\n\n def test_children_list_should_be_alphabetical_case_insensitive(self):\n with Patcher() as patcher:\n patcher.fs.create_dir(\"/data\")\n patcher.fs.create_dir(\"/data/Z\")\n patcher.fs.create_dir(\"/data/a\")\n patcher.fs.create_dir(\"/data/X\")\n patcher.fs.create_dir(\"/data/b\")\n\n self.client.force_authenticate(user=self.admin)\n response = self.client.get(\"/api/dirtree/\")\n data = response.json()[0]\n self.assertEqual(200, response.status_code)\n self.assertEqual(data[\"children\"][0][\"title\"], \"a\")\n self.assertEqual(data[\"children\"][1][\"title\"], \"b\")\n self.assertEqual(data[\"children\"][2][\"title\"], \"X\")\n self.assertEqual(data[\"children\"][3][\"title\"], \"Z\")\n\n def test_regular_user_is_not_allowed_to_retrieve_dirtree(self):\n self.client.force_authenticate(user=self.user)\n response = self.client.get(\"/api/dirtree/\")\n self.assertEqual(403, response.status_code)\n\n def test_anonymous_user_is_not_allower_to_retrieve_dirtree(self):\n self.client.force_authenticate(user=None)\n response = self.client.get(\"/api/dirtree/\")\n self.assertEqual(401, response.status_code)\n" }, { "path": "api/tests/test_duplicate_api_edge_cases.py", "content": "\"\"\"\nEdge case tests for Duplicate API to find bugs.\n\nTests cover:\n- Resolution workflow edge cases\n- Revert edge cases\n- Delete edge cases (potential Bug #13)\n- List/Detail view edge cases with missing data\n- Statistics edge cases\n\"\"\"\n\nimport uuid\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.models.duplicate import Duplicate\nfrom api.models.photo_metadata import PhotoMetadata\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass DuplicateResolveEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for duplicate resolution.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_resolve_already_resolved_duplicate(self):\n \"\"\"Test resolving a duplicate that's already resolved.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n kept_photo=photo1,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Try to resolve again with different kept photo\n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n {\"keep_photo_hash\": photo2.image_hash},\n )\n \n # Should succeed (changing which photo to keep)\n self.assertEqual(response.status_code, 200)\n \n duplicate.refresh_from_db()\n self.assertEqual(duplicate.kept_photo, photo2)\n\n def test_resolve_with_photo_already_trashed(self):\n \"\"\"Test resolving when one photo is already trashed.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user, in_trashcan=True)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Resolve keeping photo1\n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n {\"keep_photo_hash\": photo1.image_hash},\n )\n \n self.assertEqual(response.status_code, 200)\n \n # Photo2 was already trashed, shouldn't change\n photo2.refresh_from_db()\n self.assertTrue(photo2.in_trashcan)\n\n def test_resolve_with_trash_others_false(self):\n \"\"\"Test resolving without trashing other photos.\n \n Note: Must use format='json' to properly send boolean False.\n Form data converts False to string \"False\" which is truthy.\n \"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Use format='json' to properly send boolean False\n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n {\"keep_photo_hash\": photo1.image_hash, \"trash_others\": False},\n format='json',\n )\n \n self.assertEqual(response.status_code, 200)\n \n # Photo2 should NOT be trashed\n photo2.refresh_from_db()\n self.assertFalse(photo2.in_trashcan)\n \n # But duplicate should still be marked resolved\n duplicate.refresh_from_db()\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.RESOLVED)\n\n def test_resolve_with_nonexistent_photo_hash(self):\n \"\"\"Test resolving with a photo hash that doesn't exist in the group.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n {\"keep_photo_hash\": \"nonexistent_hash\"},\n )\n \n self.assertEqual(response.status_code, 400)\n self.assertIn(\"error\", response.data)\n\n def test_resolve_empty_request(self):\n \"\"\"Test resolving with empty request body.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n response = self.client.post(\n f\"/api/duplicates/{duplicate.id}/resolve/\",\n {},\n )\n \n self.assertEqual(response.status_code, 400)\n\n\nclass DuplicateRevertEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for duplicate revert.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_revert_pending_duplicate(self):\n \"\"\"Test reverting a pending (not resolved) duplicate.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n duplicate.photos.add(photo1, photo2)\n \n response = self.client.post(f\"/api/duplicates/{duplicate.id}/revert/\")\n \n # Should fail - can only revert resolved\n self.assertEqual(response.status_code, 400)\n\n def test_revert_dismissed_duplicate(self):\n \"\"\"Test reverting a dismissed duplicate.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.DISMISSED,\n )\n duplicate.photos.add(photo1, photo2)\n \n response = self.client.post(f\"/api/duplicates/{duplicate.id}/revert/\")\n \n # Should fail - can only revert resolved\n self.assertEqual(response.status_code, 400)\n\n def test_revert_when_photos_permanently_deleted(self):\n \"\"\"Test reverting when trashed photos were permanently deleted.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Resolve keeping photo1\n duplicate.resolve(kept_photo=photo1)\n \n # Now permanently delete photo2 (simulating user emptying trash)\n photo2.in_trashcan = True\n photo2.save()\n photo2.manual_delete()\n \n # Try to revert\n response = self.client.post(f\"/api/duplicates/{duplicate.id}/revert/\")\n \n # Should succeed but restored_count may be 0\n # Bug: Duplicate group might be deleted now due to Bug #12 fix\n # Let me check if duplicate still exists\n if Duplicate.objects.filter(id=duplicate.id).exists():\n self.assertEqual(response.status_code, 200)\n else:\n # Duplicate was deleted when photo2 was deleted (only 1 photo left)\n self.assertEqual(response.status_code, 404)\n\n def test_revert_multiple_times(self):\n \"\"\"Test reverting the same duplicate multiple times.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Resolve\n duplicate.resolve(kept_photo=photo1)\n \n # Revert first time\n response1 = self.client.post(f\"/api/duplicates/{duplicate.id}/revert/\")\n self.assertEqual(response1.status_code, 200)\n \n # Try to revert again (should fail - now pending)\n response2 = self.client.post(f\"/api/duplicates/{duplicate.id}/revert/\")\n self.assertEqual(response2.status_code, 400)\n\n\nclass DuplicateDeleteEdgeCasesTestCase(TestCase):\n \"\"\"\n Edge cases for duplicate delete.\n \n Note: Delete endpoint is at /api/duplicates/{id}/delete with DELETE method.\n \"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_delete_duplicate_unlinks_all_photos(self):\n \"\"\"Test that deleting a duplicate unlinks ALL photos.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(5)]\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n for photo in photos:\n duplicate.photos.add(photo)\n \n duplicate_id = duplicate.id\n \n # Correct URL: /api/duplicates/{id}/delete\n response = self.client.delete(f\"/api/duplicates/{duplicate_id}/delete\")\n \n self.assertEqual(response.status_code, 200)\n \n # Verify duplicate is deleted\n self.assertFalse(Duplicate.objects.filter(id=duplicate_id).exists())\n \n # Verify ALL photos are unlinked\n for photo in photos:\n photo.refresh_from_db()\n self.assertEqual(photo.duplicates.count(), 0,\n \"All photos should be unlinked from deleted duplicate\")\n\n def test_delete_duplicate_with_many_photos(self):\n \"\"\"Test deleting a duplicate with many photos.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(20)]\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n for photo in photos:\n duplicate.photos.add(photo)\n \n response = self.client.delete(f\"/api/duplicates/{duplicate.id}/delete\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"unlinked_count\"], 20)\n\n def test_delete_resolved_duplicate(self):\n \"\"\"Test deleting a resolved duplicate.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n kept_photo=photo1,\n )\n duplicate.photos.add(photo1, photo2)\n \n response = self.client.delete(f\"/api/duplicates/{duplicate.id}/delete\")\n \n # Should succeed - can delete any duplicate\n self.assertEqual(response.status_code, 200)\n\n def test_delete_nonexistent_duplicate(self):\n \"\"\"Test deleting a duplicate that doesn't exist.\"\"\"\n response = self.client.delete(f\"/api/duplicates/{uuid.uuid4()}/delete\")\n self.assertEqual(response.status_code, 404)\n\n def test_delete_other_users_duplicate(self):\n \"\"\"Test deleting another user's duplicate.\"\"\"\n other_user = create_test_user()\n photo1 = create_test_photo(owner=other_user)\n photo2 = create_test_photo(owner=other_user)\n \n duplicate = Duplicate.objects.create(\n owner=other_user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n response = self.client.delete(f\"/api/duplicates/{duplicate.id}/delete\")\n \n # Should return 404 (not found for this user)\n self.assertEqual(response.status_code, 404)\n\n\nclass DuplicateDetailEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for duplicate detail view.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_detail_with_photos_without_metadata(self):\n \"\"\"Test detail view when photos have no metadata.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n # Ensure no metadata exists\n PhotoMetadata.objects.filter(photo__in=[photo1, photo2]).delete()\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Should not crash\n response = self.client.get(f\"/api/duplicates/{duplicate.id}/\")\n \n self.assertEqual(response.status_code, 200)\n # Photos should have null width/height/camera\n for photo_data in response.data[\"photos\"]:\n # Should be None or return gracefully\n pass # If we get here without crash, the test passes\n\n def test_detail_with_photos_without_main_file(self):\n \"\"\"Test detail view when photos have no main_file.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo1.main_file = None\n photo1.save()\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Should not crash\n response = self.client.get(f\"/api/duplicates/{duplicate.id}/\")\n \n self.assertEqual(response.status_code, 200)\n\n def test_detail_with_deleted_kept_photo(self):\n \"\"\"Test detail view when kept_photo has been deleted.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n kept_photo=photo1,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Permanently delete the kept photo\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Check if duplicate still exists (might be deleted due to Bug #12 fix)\n if not Duplicate.objects.filter(id=duplicate.id).exists():\n # Expected behavior: duplicate deleted when < 2 photos\n return\n \n # If duplicate still exists, detail should not crash\n response = self.client.get(f\"/api/duplicates/{duplicate.id}/\")\n # Either 200 or 404 depending on photo count\n self.assertIn(response.status_code, [200, 404])\n\n\nclass DuplicateListEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for duplicate list view.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_list_excludes_single_photo_duplicates(self):\n \"\"\"Test that list excludes duplicates with only 1 photo.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1) # Only 1 photo\n \n response = self.client.get(\"/api/duplicates/\")\n \n self.assertEqual(response.status_code, 200)\n # Should not include the single-photo duplicate\n self.assertEqual(response.data[\"count\"], 0)\n\n def test_list_with_kept_photo_deleted(self):\n \"\"\"Test list view when kept_photo reference is broken.\n \n ForeignKey SET_NULL should handle this, but let's verify.\n \"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n kept_photo=photo1,\n )\n duplicate.photos.add(photo1, photo2, photo3)\n \n # Delete the kept photo\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Try listing - should not crash\n response = self.client.get(\"/api/duplicates/\")\n \n self.assertEqual(response.status_code, 200)\n\n def test_list_pagination_edge_cases(self):\n \"\"\"Test list pagination with various edge cases.\"\"\"\n # Create 5 duplicates\n for i in range(5):\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo1, photo2)\n \n # Page beyond results - Django's get_page() returns last page for out-of-range\n # So page 100 will still return results (the last page)\n response = self.client.get(\"/api/duplicates/?page=100\")\n self.assertEqual(response.status_code, 200)\n # Django returns last valid page, not empty\n self.assertGreaterEqual(len(response.data[\"results\"]), 0)\n \n # Page 0 should become page 1 (our code uses max(1, page))\n response = self.client.get(\"/api/duplicates/?page=0\")\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data[\"results\"]), 5)\n\n\nclass DuplicateAutoSelectBestEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for auto_select_best_photo.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_auto_select_with_no_photos(self):\n \"\"\"Test auto_select_best_photo with empty duplicate group.\"\"\"\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n \n result = duplicate.auto_select_best_photo()\n self.assertIsNone(result)\n\n def test_auto_select_exact_copy_all_null_paths(self):\n \"\"\"Test auto_select for exact copies when all photos have no main_file.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo1.main_file = None\n photo2.main_file = None\n photo1.save()\n photo2.save()\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Should handle gracefully (might return None or first photo)\n _result = duplicate.auto_select_best_photo()\n # Just verify it doesn't crash\n\n def test_auto_select_visual_duplicate_no_metadata(self):\n \"\"\"Test auto_select for visual duplicates when photos have no metadata.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n # Delete any metadata\n PhotoMetadata.objects.filter(photo__in=[photo1, photo2]).delete()\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Should handle gracefully\n _result = duplicate.auto_select_best_photo()\n # Just verify it doesn't crash\n\n\nclass DuplicateDismissEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for dismiss endpoint.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_dismiss_already_dismissed(self):\n \"\"\"Test dismissing an already dismissed duplicate.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.DISMISSED,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Dismiss again\n response = self.client.post(f\"/api/duplicates/{duplicate.id}/dismiss/\")\n \n # Should succeed (idempotent operation)\n self.assertEqual(response.status_code, 200)\n\n def test_dismiss_resolved_duplicate(self):\n \"\"\"Test dismissing a resolved duplicate.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n kept_photo=photo1,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Dismiss the resolved duplicate\n response = self.client.post(f\"/api/duplicates/{duplicate.id}/dismiss/\")\n \n # Should succeed\n self.assertEqual(response.status_code, 200)\n \n duplicate.refresh_from_db()\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.DISMISSED)\n\n\nclass DuplicateStatsEdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for duplicate statistics endpoint.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_stats_with_no_duplicates(self):\n \"\"\"Test stats endpoint with no duplicates.\"\"\"\n response = self.client.get(\"/api/duplicates/stats/\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"total_duplicates\"], 0)\n self.assertEqual(response.data[\"pending_duplicates\"], 0)\n self.assertEqual(response.data[\"potential_savings_bytes\"], 0)\n self.assertEqual(response.data[\"potential_savings_mb\"], 0)\n\n def test_stats_counts_by_type(self):\n \"\"\"Test stats counts by duplicate type.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n photo4 = create_test_photo(owner=self.user)\n \n # Create exact copy duplicate\n dup1 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup1.photos.add(photo1, photo2)\n \n # Create visual duplicate\n dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup2.photos.add(photo3, photo4)\n \n response = self.client.get(\"/api/duplicates/stats/\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"total_duplicates\"], 2)\n self.assertEqual(response.data[\"by_type\"][\"exact_copy\"], 1)\n self.assertEqual(response.data[\"by_type\"][\"visual_duplicate\"], 1)\n\n def test_stats_counts_by_status(self):\n \"\"\"Test stats counts by review status.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(6)]\n \n # Pending\n dup1 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n dup1.photos.add(photos[0], photos[1])\n \n # Resolved\n dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n )\n dup2.photos.add(photos[2], photos[3])\n \n # Dismissed\n dup3 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.DISMISSED,\n )\n dup3.photos.add(photos[4], photos[5])\n \n response = self.client.get(\"/api/duplicates/stats/\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"pending_duplicates\"], 1)\n self.assertEqual(response.data[\"resolved_duplicates\"], 1)\n self.assertEqual(response.data[\"dismissed_duplicates\"], 1)\n\n def test_stats_potential_savings(self):\n \"\"\"Test stats potential savings calculation.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n potential_savings=1024 * 1024 * 5, # 5 MB\n )\n dup.photos.add(photo1, photo2)\n \n response = self.client.get(\"/api/duplicates/stats/\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"potential_savings_bytes\"], 5 * 1024 * 1024)\n self.assertEqual(response.data[\"potential_savings_mb\"], 5.0)\n\n def test_stats_photos_in_duplicates(self):\n \"\"\"Test stats counts photos in duplicate groups correctly.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(5)]\n \n # Create duplicate with 3 photos\n dup1 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup1.photos.add(photos[0], photos[1], photos[2])\n \n # Create another duplicate with 2 photos (one overlapping)\n dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup2.photos.add(photos[2], photos[3]) # photos[2] is in both\n \n response = self.client.get(\"/api/duplicates/stats/\")\n \n self.assertEqual(response.status_code, 200)\n # 4 unique photos are in duplicate groups (photos 0,1,2,3)\n self.assertEqual(response.data[\"photos_in_duplicates\"], 4)\n\n def test_stats_other_users_not_included(self):\n \"\"\"Test stats don't include other user's duplicates.\"\"\"\n other_user = create_test_user()\n \n # Create duplicate for other user\n other_photo1 = create_test_photo(owner=other_user)\n other_photo2 = create_test_photo(owner=other_user)\n dup = Duplicate.objects.create(\n owner=other_user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(other_photo1, other_photo2)\n \n # Current user should see 0 duplicates\n response = self.client.get(\"/api/duplicates/stats/\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"total_duplicates\"], 0)\n" }, { "path": "api/tests/test_duplicate_detection.py", "content": "\"\"\"\nTests for Duplicate Detection API and Models.\n\nTests cover:\n- Duplicate model creation, resolution, dismissal, revert\n- API endpoints for listing, filtering, resolving duplicates\n- Edge cases: permissions, invalid IDs, concurrent operations\n- BK-Tree algorithm for visual duplicate search\n\"\"\"\nimport uuid\n\nfrom django.test import TestCase\nfrom rest_framework import status\nfrom rest_framework.test import APIClient\n\nfrom api.models.duplicate import Duplicate\nfrom api.tests.utils import create_test_photos, create_test_user\n\n\nclass DuplicateModelTest(TestCase):\n \"\"\"Tests for the Duplicate model methods.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.photos = create_test_photos(number_of_photos=3, owner=self.user)\n\n def test_create_duplicate_group(self):\n \"\"\"Test creating a duplicate group with multiple photos.\"\"\"\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n for photo in self.photos:\n photo.duplicates.add(duplicate)\n\n self.assertEqual(duplicate.photo_count, 3)\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.PENDING)\n\n def test_create_duplicate_with_less_than_2_photos_returns_none(self):\n \"\"\"Test create_or_merge returns None with < 2 photos.\"\"\"\n result = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=[self.photos[0]],\n )\n self.assertIsNone(result)\n\n def test_create_or_merge_creates_new_duplicate(self):\n \"\"\"Test create_or_merge creates a new duplicate group.\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n photos=self.photos[:2],\n similarity_score=0.95,\n )\n self.assertIsNotNone(duplicate)\n self.assertEqual(duplicate.photo_count, 2)\n self.assertEqual(duplicate.similarity_score, 0.95)\n\n def test_create_or_merge_merges_existing(self):\n \"\"\"Test create_or_merge merges when photo already in group.\"\"\"\n # Create initial group with first 2 photos\n dup1 = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n # Try to create new group with overlapping photo\n dup2 = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=[self.photos[1], self.photos[2]],\n )\n # Should return the same duplicate (merged)\n self.assertEqual(dup1.id, dup2.id)\n self.assertEqual(dup1.photo_count, 3)\n\n def test_resolve_duplicate_trashes_others(self):\n \"\"\"Test resolving a duplicate trashes non-kept photos.\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos,\n )\n keep_photo = self.photos[0]\n duplicate.resolve(keep_photo, trash_others=True)\n\n # Refresh from DB\n duplicate.refresh_from_db()\n for photo in self.photos[1:]:\n photo.refresh_from_db()\n self.assertTrue(photo.in_trashcan)\n\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.RESOLVED)\n self.assertEqual(duplicate.kept_photo, keep_photo)\n self.assertEqual(duplicate.trashed_count, 2)\n\n def test_resolve_duplicate_without_trashing(self):\n \"\"\"Test resolving without trashing others.\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n keep_photo = self.photos[0]\n duplicate.resolve(keep_photo, trash_others=False)\n\n duplicate.refresh_from_db()\n self.photos[1].refresh_from_db()\n\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.RESOLVED)\n self.assertFalse(self.photos[1].in_trashcan)\n self.assertEqual(duplicate.trashed_count, 0)\n\n def test_dismiss_duplicate(self):\n \"\"\"Test dismissing a duplicate unlinks photos.\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n photos=self.photos[:2],\n )\n duplicate.dismiss()\n\n duplicate.refresh_from_db()\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.DISMISSED)\n # Photos should be unlinked\n self.assertEqual(duplicate.photo_count, 0)\n\n def test_revert_resolved_duplicate(self):\n \"\"\"Test reverting a resolved duplicate restores photos.\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n duplicate.resolve(self.photos[0], trash_others=True)\n\n # Verify photo was trashed\n self.photos[1].refresh_from_db()\n self.assertTrue(self.photos[1].in_trashcan)\n\n # Revert\n restored_count = duplicate.revert()\n duplicate.refresh_from_db()\n self.photos[1].refresh_from_db()\n\n self.assertEqual(restored_count, 1)\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.PENDING)\n self.assertFalse(self.photos[1].in_trashcan)\n self.assertIsNone(duplicate.kept_photo)\n\n def test_revert_non_resolved_duplicate_returns_zero(self):\n \"\"\"Test reverting a pending duplicate returns 0.\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n restored_count = duplicate.revert()\n self.assertEqual(restored_count, 0)\n\n def test_auto_select_best_photo_exact_copy(self):\n \"\"\"Test auto-selecting best photo for exact copies (shortest path).\"\"\"\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n best = duplicate.auto_select_best_photo()\n self.assertIsNotNone(best)\n\n def test_calculate_potential_savings(self):\n \"\"\"Test calculating potential storage savings.\"\"\"\n # Set known sizes\n self.photos[0].size = 1000000 # 1MB\n self.photos[0].save()\n self.photos[1].size = 500000 # 0.5MB\n self.photos[1].save()\n\n duplicate = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n savings = duplicate.calculate_potential_savings()\n\n # Should be size of non-best photos\n self.assertGreater(savings, 0)\n\n def test_merge_duplicates(self):\n \"\"\"Test merging two duplicate groups.\"\"\"\n dup1 = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n # Create second group with photo[2]\n extra_photo = create_test_photos(number_of_photos=1, owner=self.user)[0]\n dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n extra_photo.duplicates.add(dup2)\n self.photos[2].duplicates.add(dup2)\n\n # Merge\n dup1.merge_with(dup2)\n\n # dup2 should be deleted\n self.assertFalse(Duplicate.objects.filter(id=dup2.id).exists())\n # dup1 should have all photos\n self.assertEqual(dup1.photo_count, 4)\n\n\nclass DuplicateAPITest(TestCase):\n \"\"\"Tests for Duplicate API endpoints.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n self.photos = create_test_photos(number_of_photos=4, owner=self.user1)\n\n def test_list_duplicates_empty(self):\n \"\"\"Test listing duplicates when none exist.\"\"\"\n response = self.client.get(\"/api/duplicates\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"count\"], 0)\n self.assertEqual(data[\"results\"], [])\n\n def test_list_duplicates_with_results(self):\n \"\"\"Test listing duplicates with results.\"\"\"\n Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n response = self.client.get(\"/api/duplicates\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"count\"], 1)\n\n def test_list_duplicates_excludes_other_users(self):\n \"\"\"Test that user can only see their own duplicates.\"\"\"\n # Create duplicate for user2\n photos2 = create_test_photos(number_of_photos=2, owner=self.user2)\n Duplicate.create_or_merge(\n owner=self.user2,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=photos2,\n )\n # User1 should not see it\n response = self.client.get(\"/api/duplicates\")\n data = response.json()\n self.assertEqual(data[\"count\"], 0)\n\n def test_list_duplicates_filter_by_type(self):\n \"\"\"Test filtering duplicates by type.\"\"\"\n Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n photos=self.photos[2:4],\n )\n\n # Filter exact copies\n response = self.client.get(\"/api/duplicates?duplicate_type=exact_copy\")\n data = response.json()\n self.assertEqual(data[\"count\"], 1)\n self.assertEqual(data[\"results\"][0][\"duplicate_type\"], \"exact_copy\")\n\n def test_list_duplicates_filter_by_status(self):\n \"\"\"Test filtering duplicates by review status.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n dup.resolve(self.photos[0], trash_others=True)\n\n # Filter pending - should be empty\n response = self.client.get(\"/api/duplicates?status=pending\")\n data = response.json()\n self.assertEqual(data[\"count\"], 0)\n\n # Filter resolved\n response = self.client.get(\"/api/duplicates?status=resolved\")\n data = response.json()\n self.assertEqual(data[\"count\"], 1)\n\n def test_get_duplicate_detail(self):\n \"\"\"Test getting duplicate detail.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n response = self.client.get(f\"/api/duplicates/{dup.id}\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"id\"], str(dup.id))\n self.assertEqual(len(data[\"photos\"]), 2)\n\n def test_get_duplicate_detail_not_found(self):\n \"\"\"Test getting non-existent duplicate returns 404.\"\"\"\n fake_id = uuid.uuid4()\n response = self.client.get(f\"/api/duplicates/{fake_id}\")\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_get_duplicate_detail_wrong_user(self):\n \"\"\"Test user cannot access other user's duplicate.\"\"\"\n photos2 = create_test_photos(number_of_photos=2, owner=self.user2)\n dup = Duplicate.create_or_merge(\n owner=self.user2,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=photos2,\n )\n response = self.client.get(f\"/api/duplicates/{dup.id}\")\n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_resolve_duplicate(self):\n \"\"\"Test resolving a duplicate via API.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n payload = {\n \"keep_photo_hash\": self.photos[0].image_hash,\n \"trash_others\": True,\n }\n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n data=payload,\n format=\"json\",\n )\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"status\"], \"resolved\")\n self.assertEqual(data[\"trashed_count\"], 1)\n\n def test_resolve_duplicate_missing_photo_hash(self):\n \"\"\"Test resolve without photo hash returns error.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve/\",\n data={},\n format=\"json\",\n )\n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_resolve_duplicate_invalid_photo(self):\n \"\"\"Test resolve with photo not in group returns error.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n payload = {\n \"keep_photo_hash\": self.photos[3].image_hash, # Not in group\n \"trash_others\": True,\n }\n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n data=payload,\n format=\"json\",\n )\n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_dismiss_duplicate(self):\n \"\"\"Test dismissing a duplicate via API.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n photos=self.photos[:2],\n )\n response = self.client.post(f\"/api/duplicates/{dup.id}/dismiss\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"status\"], \"dismissed\")\n\n def test_revert_duplicate(self):\n \"\"\"Test reverting a resolved duplicate via API.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n dup.resolve(self.photos[0], trash_others=True)\n\n response = self.client.post(f\"/api/duplicates/{dup.id}/revert\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"status\"], \"reverted\")\n self.assertEqual(data[\"restored_count\"], 1)\n\n def test_revert_non_resolved_duplicate_fails(self):\n \"\"\"Test reverting a pending duplicate returns error.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n response = self.client.post(f\"/api/duplicates/{dup.id}/revert\")\n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_delete_duplicate(self):\n \"\"\"Test deleting a duplicate group via API.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n dup_id = dup.id\n # API uses /delete suffix instead of DELETE method on main path\n response = self.client.delete(f\"/api/duplicates/{dup_id}/delete\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"status\"], \"deleted\")\n self.assertEqual(data[\"unlinked_count\"], 2)\n\n # Verify duplicate is gone\n self.assertFalse(Duplicate.objects.filter(id=dup_id).exists())\n\n def test_get_duplicate_stats(self):\n \"\"\"Test getting duplicate statistics.\"\"\"\n # Create some duplicates\n Duplicate.create_or_merge(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n response = self.client.get(\"/api/duplicates/stats\")\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertIn(\"total_duplicates\", data)\n self.assertIn(\"pending_duplicates\", data)\n self.assertIn(\"by_type\", data)\n\n def test_detect_duplicates(self):\n \"\"\"Test triggering duplicate detection.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n data={\n \"detect_exact_copies\": True,\n \"detect_visual_duplicates\": False,\n },\n format=\"json\",\n )\n self.assertEqual(response.status_code, status.HTTP_202_ACCEPTED)\n data = response.json()\n self.assertEqual(data[\"status\"], \"queued\")\n\n\nclass DuplicateEdgeCasesTest(TestCase):\n \"\"\"Tests for edge cases and potential bugs.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n self.photos = create_test_photos(number_of_photos=5, owner=self.user)\n\n def test_duplicate_with_single_photo_excluded_from_list(self):\n \"\"\"Test duplicates with <2 photos are not returned in list.\"\"\"\n # Create a duplicate and manually remove all but one photo\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n self.photos[0].duplicates.add(dup) # Only 1 photo\n\n response = self.client.get(\"/api/duplicates/\")\n data = response.json()\n self.assertEqual(data[\"count\"], 0)\n\n def test_resolve_already_resolved_duplicate(self):\n \"\"\"Test resolving an already resolved duplicate.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:3],\n )\n dup.resolve(self.photos[0], trash_others=True)\n\n # Try to resolve again with different photo\n payload = {\"keep_photo_hash\": self.photos[1].image_hash}\n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n data=payload,\n format=\"json\",\n )\n # Should still succeed (updating the resolution)\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n\n def test_photo_in_multiple_duplicate_groups(self):\n \"\"\"Test a photo can be in multiple duplicate groups of different types.\"\"\"\n _dup_exact = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:2],\n )\n _dup_visual = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n photos=[self.photos[0], self.photos[2]],\n )\n\n # Photo 0 should be in both groups\n self.photos[0].refresh_from_db()\n self.assertEqual(self.photos[0].duplicates.count(), 2)\n\n def test_delete_photo_removes_from_duplicate(self):\n \"\"\"Test deleting a photo removes it from duplicate group.\"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:3],\n )\n initial_count = dup.photo_count\n\n # Delete a photo from the duplicate group\n photo_to_delete = self.photos[0]\n photo_to_delete.duplicates.remove(dup)\n\n dup.refresh_from_db()\n self.assertEqual(dup.photo_count, initial_count - 1)\n\n def test_pagination_works_correctly(self):\n \"\"\"Test pagination returns correct results.\"\"\"\n # Create 25 duplicate groups\n for i in range(25):\n extra_photos = create_test_photos(number_of_photos=2, owner=self.user)\n Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=extra_photos,\n )\n\n # Get first page\n response = self.client.get(\"/api/duplicates?page=1&page_size=10\")\n data = response.json()\n self.assertEqual(len(data[\"results\"]), 10)\n self.assertEqual(data[\"count\"], 25)\n self.assertTrue(data[\"has_next\"])\n\n # Get second page\n response = self.client.get(\"/api/duplicates?page=2&page_size=10\")\n data = response.json()\n self.assertEqual(len(data[\"results\"]), 10)\n self.assertTrue(data[\"has_previous\"])\n\n def test_invalid_uuid_format(self):\n \"\"\"Test invalid UUID format.\n \n Note: The URL regex pattern [0-9a-f-]+ matches any hex-like string,\n so 'not-a-valid-uuid' partially matches. The view then returns an\n empty result rather than 404/400. This is acceptable behavior.\n \"\"\"\n response = self.client.get(\"/api/duplicates/not-a-valid-uuid\")\n # The regex matches the string (contains a-f and -), but no duplicate exists\n # so it returns the list endpoint with empty results\n self.assertEqual(response.status_code, status.HTTP_200_OK)\n\n def test_concurrent_resolve_same_duplicate(self):\n \"\"\"Test handling multiple resolution attempts.\n \n The API allows re-resolving a duplicate with the same or different photo.\n This is useful if the user changes their mind about which photo to keep.\n \"\"\"\n dup = Duplicate.create_or_merge(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n photos=self.photos[:3],\n )\n # First resolve - keep photos[0], trash others\n payload1 = {\"keep_photo_hash\": self.photos[0].image_hash}\n response1 = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n data=payload1,\n format=\"json\",\n )\n self.assertEqual(response1.status_code, status.HTTP_200_OK)\n\n # Verify photos[1] is now trashed\n self.photos[1].refresh_from_db()\n self.assertTrue(self.photos[1].in_trashcan)\n\n # Re-resolve with same photo (idempotent operation)\n payload2 = {\"keep_photo_hash\": self.photos[0].image_hash}\n response2 = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n data=payload2,\n format=\"json\",\n )\n # Re-resolving is allowed\n self.assertEqual(response2.status_code, status.HTTP_200_OK)\n\n\nclass BKTreeTest(TestCase):\n \"\"\"Tests for BK-Tree algorithm used in visual duplicate detection.\"\"\"\n\n def test_bk_tree_basic_operations(self):\n \"\"\"Test BK-Tree add and search operations.\"\"\"\n from api.duplicate_detection import BKTree\n from api.perceptual_hash import hamming_distance\n\n tree = BKTree(hamming_distance)\n tree.add(\"photo1\", \"0000000000000000\")\n tree.add(\"photo2\", \"0000000000000001\") # 1 bit different\n tree.add(\"photo3\", \"1111111111111111\") # Very different\n\n # Search with threshold 1\n results = tree.search(\"0000000000000000\", 1)\n result_ids = [r[0] for r in results]\n self.assertIn(\"photo1\", result_ids)\n self.assertIn(\"photo2\", result_ids)\n self.assertNotIn(\"photo3\", result_ids)\n\n def test_bk_tree_empty_search(self):\n \"\"\"Test BK-Tree search on empty tree.\"\"\"\n from api.duplicate_detection import BKTree\n from api.perceptual_hash import hamming_distance\n\n tree = BKTree(hamming_distance)\n results = tree.search(\"0000000000000000\", 5)\n self.assertEqual(results, [])\n\n\nclass UnionFindTest(TestCase):\n \"\"\"Tests for Union-Find data structure used in duplicate grouping.\"\"\"\n\n def test_union_find_basic(self):\n \"\"\"Test Union-Find basic operations.\"\"\"\n from api.duplicate_detection import UnionFind\n\n uf = UnionFind()\n uf.union(\"a\", \"b\")\n uf.union(\"b\", \"c\")\n\n # a, b, c should be in same group\n self.assertEqual(uf.find(\"a\"), uf.find(\"b\"))\n self.assertEqual(uf.find(\"b\"), uf.find(\"c\"))\n\n def test_union_find_get_groups(self):\n \"\"\"Test Union-Find get_groups returns correct groups.\"\"\"\n from api.duplicate_detection import UnionFind\n\n uf = UnionFind()\n uf.union(\"a\", \"b\")\n uf.union(\"c\", \"d\")\n uf.union(\"e\", \"f\")\n\n groups = uf.get_groups()\n self.assertEqual(len(groups), 3)\n # Each group should have 2 elements\n for group in groups:\n self.assertEqual(len(group), 2)\n\n def test_union_find_single_elements_not_returned(self):\n \"\"\"Test Union-Find doesn't return single-element groups.\"\"\"\n from api.duplicate_detection import UnionFind\n\n uf = UnionFind()\n uf.find(\"a\") # Creates single element\n uf.union(\"b\", \"c\")\n\n groups = uf.get_groups()\n # Only the b-c group should be returned\n self.assertEqual(len(groups), 1)\n self.assertEqual(len(groups[0]), 2)\n" }, { "path": "api/tests/test_duplicate_detection_logic.py", "content": "\"\"\"\nComprehensive tests for Duplicate Detection Logic.\n\nTests cover:\n- BKTree: Burkhard-Keller Tree for efficient Hamming distance queries\n- UnionFind: Union-Find data structure for grouping\n- detect_exact_copies: Exact copy detection via MD5 hash\n- detect_visual_duplicates: Visual similarity detection via perceptual hash\n- batch_detect_duplicates: Batch detection orchestration\n- Edge cases and error handling\n\"\"\"\n\nimport uuid\nfrom unittest.mock import patch\n\nfrom django.test import TestCase\n\nfrom api.models.duplicate import Duplicate\nfrom api.models.file import File\nfrom api.models.long_running_job import LongRunningJob\nfrom api.duplicate_detection import (\n BKTree,\n UnionFind,\n detect_exact_copies,\n detect_visual_duplicates,\n batch_detect_duplicates,\n)\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass BKTreeTestCase(TestCase):\n \"\"\"Tests for BKTree data structure.\"\"\"\n\n def setUp(self):\n # Simple Hamming distance for testing\n def hamming(a, b):\n return sum(c1 != c2 for c1, c2 in zip(a, b))\n self.tree = BKTree(hamming)\n\n def test_empty_tree_search_returns_empty(self):\n \"\"\"Test searching empty tree returns empty list.\"\"\"\n results = self.tree.search(\"abc\", 1)\n self.assertEqual(results, [])\n\n def test_add_single_item(self):\n \"\"\"Test adding single item to tree.\"\"\"\n self.tree.add(\"id1\", \"abc\")\n \n self.assertEqual(self.tree.size, 1)\n self.assertIsNotNone(self.tree.root)\n self.assertEqual(self.tree.root[\"id\"], \"id1\")\n self.assertEqual(self.tree.root[\"hash\"], \"abc\")\n\n def test_add_multiple_items(self):\n \"\"\"Test adding multiple items to tree.\"\"\"\n self.tree.add(\"id1\", \"abc\")\n self.tree.add(\"id2\", \"abd\")\n self.tree.add(\"id3\", \"xyz\")\n \n self.assertEqual(self.tree.size, 3)\n\n def test_search_exact_match(self):\n \"\"\"Test searching for exact match (distance 0).\"\"\"\n self.tree.add(\"id1\", \"abc\")\n self.tree.add(\"id2\", \"xyz\")\n \n results = self.tree.search(\"abc\", 0)\n \n self.assertEqual(len(results), 1)\n self.assertEqual(results[0][0], \"id1\")\n self.assertEqual(results[0][1], 0) # distance\n\n def test_search_within_threshold(self):\n \"\"\"Test searching within Hamming threshold.\"\"\"\n self.tree.add(\"id1\", \"abc\")\n self.tree.add(\"id2\", \"abd\") # distance 1 from \"abc\"\n self.tree.add(\"id3\", \"xyz\") # distance 3 from \"abc\"\n \n results = self.tree.search(\"abc\", 1)\n \n self.assertEqual(len(results), 2)\n result_ids = [r[0] for r in results]\n self.assertIn(\"id1\", result_ids)\n self.assertIn(\"id2\", result_ids)\n self.assertNotIn(\"id3\", result_ids)\n\n def test_search_threshold_excludes_distant(self):\n \"\"\"Test threshold correctly excludes distant items.\"\"\"\n self.tree.add(\"id1\", \"aaaa\")\n self.tree.add(\"id2\", \"zzzz\") # distance 4 from \"aaaa\"\n \n results = self.tree.search(\"aaaa\", 3)\n \n self.assertEqual(len(results), 1)\n self.assertEqual(results[0][0], \"id1\")\n\n def test_search_returns_correct_distances(self):\n \"\"\"Test search returns correct Hamming distances.\"\"\"\n self.tree.add(\"id1\", \"abc\")\n self.tree.add(\"id2\", \"aXc\") # distance 1\n self.tree.add(\"id3\", \"XXc\") # distance 2\n \n results = self.tree.search(\"abc\", 2)\n \n results_dict = {r[0]: r[1] for r in results}\n self.assertEqual(results_dict[\"id1\"], 0)\n self.assertEqual(results_dict[\"id2\"], 1)\n self.assertEqual(results_dict[\"id3\"], 2)\n\n def test_add_duplicate_hash(self):\n \"\"\"Test adding items with same hash.\"\"\"\n self.tree.add(\"id1\", \"abc\")\n self.tree.add(\"id2\", \"abc\") # Same hash, different id\n \n self.assertEqual(self.tree.size, 2)\n \n results = self.tree.search(\"abc\", 0)\n result_ids = [r[0] for r in results]\n self.assertIn(\"id1\", result_ids)\n self.assertIn(\"id2\", result_ids)\n\n def test_large_tree_performance(self):\n \"\"\"Test tree performs well with many items.\"\"\"\n # Add 1000 items\n for i in range(1000):\n hash_val = f\"{i:04d}\"\n self.tree.add(f\"id{i}\", hash_val)\n \n self.assertEqual(self.tree.size, 1000)\n \n # Search should complete quickly\n results = self.tree.search(\"0500\", 1)\n self.assertGreater(len(results), 0)\n\n\nclass UnionFindTestCase(TestCase):\n \"\"\"Tests for UnionFind data structure.\"\"\"\n\n def test_initial_find_creates_entry(self):\n \"\"\"Test find creates new entry if not exists.\"\"\"\n uf = UnionFind()\n \n root = uf.find(\"a\")\n \n self.assertEqual(root, \"a\")\n self.assertIn(\"a\", uf.parent)\n\n def test_find_same_element_returns_itself(self):\n \"\"\"Test find on single element returns itself.\"\"\"\n uf = UnionFind()\n \n uf.find(\"a\")\n root = uf.find(\"a\")\n \n self.assertEqual(root, \"a\")\n\n def test_union_links_elements(self):\n \"\"\"Test union links two elements.\"\"\"\n uf = UnionFind()\n \n uf.union(\"a\", \"b\")\n \n self.assertEqual(uf.find(\"a\"), uf.find(\"b\"))\n\n def test_union_multiple_elements(self):\n \"\"\"Test union of multiple elements.\"\"\"\n uf = UnionFind()\n \n uf.union(\"a\", \"b\")\n uf.union(\"b\", \"c\")\n uf.union(\"c\", \"d\")\n \n # All should have same root\n root_a = uf.find(\"a\")\n self.assertEqual(uf.find(\"b\"), root_a)\n self.assertEqual(uf.find(\"c\"), root_a)\n self.assertEqual(uf.find(\"d\"), root_a)\n\n def test_union_separate_groups(self):\n \"\"\"Test union keeps separate groups separate.\"\"\"\n uf = UnionFind()\n \n uf.union(\"a\", \"b\")\n uf.union(\"c\", \"d\")\n \n self.assertEqual(uf.find(\"a\"), uf.find(\"b\"))\n self.assertEqual(uf.find(\"c\"), uf.find(\"d\"))\n self.assertNotEqual(uf.find(\"a\"), uf.find(\"c\"))\n\n def test_get_groups_returns_groups(self):\n \"\"\"Test get_groups returns correct groups.\"\"\"\n uf = UnionFind()\n \n uf.union(\"a\", \"b\")\n uf.union(\"c\", \"d\")\n uf.union(\"d\", \"e\")\n \n groups = uf.get_groups()\n \n self.assertEqual(len(groups), 2)\n \n # Check group contents\n group_sets = [set(g) for g in groups]\n self.assertIn({\"a\", \"b\"}, group_sets)\n self.assertIn({\"c\", \"d\", \"e\"}, group_sets)\n\n def test_get_groups_excludes_singletons(self):\n \"\"\"Test get_groups excludes single-element groups.\"\"\"\n uf = UnionFind()\n \n uf.find(\"a\") # Single element\n uf.union(\"b\", \"c\") # Pair\n \n groups = uf.get_groups()\n \n self.assertEqual(len(groups), 1)\n self.assertEqual(set(groups[0]), {\"b\", \"c\"})\n\n def test_path_compression(self):\n \"\"\"Test path compression works (parent points directly to root).\"\"\"\n uf = UnionFind()\n \n # Create long chain\n uf.union(\"a\", \"b\")\n uf.union(\"b\", \"c\")\n uf.union(\"c\", \"d\")\n \n # After find, path should be compressed\n root = uf.find(\"a\")\n self.assertEqual(uf.parent[\"a\"], root)\n\n\nclass DetectExactCopiesTestCase(TestCase):\n \"\"\"Tests for detect_exact_copies function.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def _create_photo_with_hash(self, image_hash, file_hash=None, **kwargs):\n \"\"\"Helper to create Photo with specific hashes.\"\"\"\n photo = create_test_photo(owner=self.user, **kwargs)\n photo.image_hash = image_hash\n \n if file_hash:\n file = File.objects.create(\n hash=file_hash,\n path=f\"/photos/test_{uuid.uuid4()}.jpg\",\n type=File.IMAGE,\n )\n photo.files.add(file)\n photo.main_file = file\n \n photo.save()\n return photo\n\n def test_no_duplicates_returns_zero(self):\n \"\"\"Test no duplicates when all hashes unique.\"\"\"\n self._create_photo_with_hash(\"hash1\")\n self._create_photo_with_hash(\"hash2\")\n self._create_photo_with_hash(\"hash3\")\n \n count = detect_exact_copies(self.user)\n \n self.assertEqual(count, 0)\n\n def test_detects_duplicate_image_hash(self):\n \"\"\"Test detects photos with same image_hash.\"\"\"\n self._create_photo_with_hash(\"same_hash\")\n self._create_photo_with_hash(\"same_hash\")\n \n count = detect_exact_copies(self.user)\n \n self.assertEqual(count, 1)\n \n # Check duplicate was created\n duplicates = Duplicate.objects.filter(owner=self.user)\n self.assertEqual(duplicates.count(), 1)\n self.assertEqual(duplicates.first().photos.count(), 2)\n self.assertEqual(duplicates.first().duplicate_type, Duplicate.DuplicateType.EXACT_COPY)\n\n def test_detects_duplicate_file_hash(self):\n \"\"\"Test detects photos with same file hash (MD5 part).\"\"\"\n # Same MD5 content hash (first 32 chars)\n file_hash1 = \"a\" * 32 + \"user1\"\n file_hash2 = \"a\" * 32 + \"user2\" # Same MD5, different suffix\n \n self._create_photo_with_hash(\"unique1\", file_hash1)\n self._create_photo_with_hash(\"unique2\", file_hash2)\n \n count = detect_exact_copies(self.user)\n \n self.assertEqual(count, 1)\n\n def test_skips_hidden_photos(self):\n \"\"\"Test hidden photos are excluded.\"\"\"\n self._create_photo_with_hash(\"same_hash\", hidden=True)\n self._create_photo_with_hash(\"same_hash\")\n \n count = detect_exact_copies(self.user)\n \n # Only 1 visible photo with this hash, so no duplicate\n self.assertEqual(count, 0)\n\n def test_skips_trashed_photos(self):\n \"\"\"Test trashed photos are excluded.\"\"\"\n self._create_photo_with_hash(\"same_hash\", in_trashcan=True)\n self._create_photo_with_hash(\"same_hash\")\n \n count = detect_exact_copies(self.user)\n \n self.assertEqual(count, 0)\n\n def test_multiple_duplicate_groups(self):\n \"\"\"Test detects multiple separate duplicate groups.\"\"\"\n # Group 1\n self._create_photo_with_hash(\"hash_a\")\n self._create_photo_with_hash(\"hash_a\")\n \n # Group 2\n self._create_photo_with_hash(\"hash_b\")\n self._create_photo_with_hash(\"hash_b\")\n self._create_photo_with_hash(\"hash_b\")\n \n count = detect_exact_copies(self.user)\n \n self.assertEqual(count, 2)\n\n def test_progress_callback_called(self):\n \"\"\"Test progress callback is called during detection.\"\"\"\n self._create_photo_with_hash(\"same_hash\")\n self._create_photo_with_hash(\"same_hash\")\n \n callback_calls = []\n \n def progress_callback(current, total, found):\n callback_calls.append((current, total, found))\n \n detect_exact_copies(self.user, progress_callback=progress_callback)\n \n # Callback may or may not be called depending on group count\n\n\nclass DetectVisualDuplicatesTestCase(TestCase):\n \"\"\"Tests for detect_visual_duplicates function.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def _create_photo_with_phash(self, perceptual_hash, **kwargs):\n \"\"\"Helper to create Photo with perceptual hash.\"\"\"\n photo = create_test_photo(owner=self.user, **kwargs)\n photo.perceptual_hash = perceptual_hash\n photo.save()\n return photo\n\n def test_no_photos_returns_zero(self):\n \"\"\"Test returns 0 when no photos.\"\"\"\n count = detect_visual_duplicates(self.user)\n \n self.assertEqual(count, 0)\n\n def test_single_photo_returns_zero(self):\n \"\"\"Test returns 0 with only one photo.\"\"\"\n self._create_photo_with_phash(\"abcd1234\")\n \n count = detect_visual_duplicates(self.user)\n \n self.assertEqual(count, 0)\n\n def test_detects_identical_phash(self):\n \"\"\"Test detects photos with identical perceptual hash.\"\"\"\n self._create_photo_with_phash(\"a\" * 16)\n self._create_photo_with_phash(\"a\" * 16)\n \n count = detect_visual_duplicates(self.user, threshold=0)\n \n self.assertEqual(count, 1)\n\n def test_detects_similar_phash_within_threshold(self):\n \"\"\"Test detects photos with similar perceptual hash.\"\"\"\n # These hashes differ by 2 characters\n self._create_photo_with_phash(\"aaaaaaaaaaaaaaaa\")\n self._create_photo_with_phash(\"aaaaaaaaaaaaaabb\") # 2 chars different\n \n count = detect_visual_duplicates(self.user, threshold=5)\n \n self.assertEqual(count, 1)\n\n def test_threshold_excludes_dissimilar(self):\n \"\"\"Test threshold excludes dissimilar photos.\"\"\"\n self._create_photo_with_phash(\"aaaaaaaaaaaaaaaa\")\n self._create_photo_with_phash(\"zzzzzzzzzzzzzzzz\") # Very different\n \n count = detect_visual_duplicates(self.user, threshold=5)\n \n self.assertEqual(count, 0)\n\n def test_skips_photos_without_phash(self):\n \"\"\"Test photos without perceptual hash are skipped.\"\"\"\n self._create_photo_with_phash(\"aaaaaaaaaaaaaaaa\")\n photo_no_hash = create_test_photo(owner=self.user)\n photo_no_hash.perceptual_hash = None\n photo_no_hash.save()\n \n count = detect_visual_duplicates(self.user)\n \n self.assertEqual(count, 0)\n\n def test_skips_hidden_photos(self):\n \"\"\"Test hidden photos are excluded.\"\"\"\n self._create_photo_with_phash(\"aaaaaaaaaaaaaaaa\", hidden=True)\n self._create_photo_with_phash(\"aaaaaaaaaaaaaaaa\")\n \n count = detect_visual_duplicates(self.user, threshold=0)\n \n self.assertEqual(count, 0)\n\n def test_creates_visual_duplicate_type(self):\n \"\"\"Test creates duplicate with VISUAL_DUPLICATE type.\"\"\"\n self._create_photo_with_phash(\"a\" * 16)\n self._create_photo_with_phash(\"a\" * 16)\n \n detect_visual_duplicates(self.user, threshold=0)\n \n duplicate = Duplicate.objects.filter(owner=self.user).first()\n self.assertEqual(duplicate.duplicate_type, Duplicate.DuplicateType.VISUAL_DUPLICATE)\n\n\nclass BatchDetectDuplicatesTestCase(TestCase):\n \"\"\"Tests for batch_detect_duplicates orchestration.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n @patch('api.duplicate_detection.detect_exact_copies')\n @patch('api.duplicate_detection.detect_visual_duplicates')\n def test_calls_both_detectors_by_default(self, mock_visual, mock_exact):\n \"\"\"Test both detectors called with default options.\"\"\"\n mock_exact.return_value = 5\n mock_visual.return_value = 3\n \n batch_detect_duplicates(self.user)\n \n mock_exact.assert_called_once()\n mock_visual.assert_called_once()\n\n @patch('api.duplicate_detection.detect_exact_copies')\n @patch('api.duplicate_detection.detect_visual_duplicates')\n def test_respects_options(self, mock_visual, mock_exact):\n \"\"\"Test options control which detectors run.\"\"\"\n mock_exact.return_value = 0\n mock_visual.return_value = 0\n \n batch_detect_duplicates(self.user, options={\n 'detect_exact_copies': False,\n 'detect_visual_duplicates': True,\n })\n \n mock_exact.assert_not_called()\n mock_visual.assert_called_once()\n\n @patch('api.duplicate_detection.detect_exact_copies')\n @patch('api.duplicate_detection.detect_visual_duplicates')\n def test_passes_visual_threshold(self, mock_visual, mock_exact):\n \"\"\"Test visual threshold passed to detector.\"\"\"\n mock_exact.return_value = 0\n mock_visual.return_value = 0\n \n batch_detect_duplicates(self.user, options={\n 'visual_threshold': 15,\n })\n \n mock_visual.assert_called_once()\n args, kwargs = mock_visual.call_args\n self.assertEqual(args[1], 15) # threshold argument\n\n @patch('api.duplicate_detection.detect_exact_copies')\n @patch('api.duplicate_detection.detect_visual_duplicates')\n def test_creates_job(self, mock_visual, mock_exact):\n \"\"\"Test LongRunningJob created for tracking.\"\"\"\n mock_exact.return_value = 0\n mock_visual.return_value = 0\n \n batch_detect_duplicates(self.user)\n \n job = LongRunningJob.objects.filter(\n started_by=self.user,\n job_type=LongRunningJob.JOB_DETECT_DUPLICATES,\n ).first()\n self.assertIsNotNone(job)\n\n @patch('api.duplicate_detection.detect_exact_copies')\n @patch('api.duplicate_detection.detect_visual_duplicates')\n def test_clear_pending_option(self, mock_visual, mock_exact):\n \"\"\"Test clear_pending option clears pending duplicates.\"\"\"\n mock_exact.return_value = 0\n mock_visual.return_value = 0\n \n # Create pending duplicate\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n duplicate.photos.add(photo1, photo2)\n \n batch_detect_duplicates(self.user, options={'clear_pending': True})\n \n # Pending duplicate should be cleared\n self.assertFalse(Duplicate.objects.filter(pk=duplicate.pk).exists())\n\n @patch('api.duplicate_detection.detect_exact_copies')\n def test_handles_exception(self, mock_exact):\n \"\"\"Test exception handling during detection.\"\"\"\n mock_exact.side_effect = Exception(\"Detection failed\")\n \n with self.assertRaises(Exception):\n batch_detect_duplicates(self.user)\n \n # Job should be marked as failed\n job = LongRunningJob.objects.filter(started_by=self.user).first()\n self.assertIsNotNone(job)\n\n\nclass EdgeCasesTestCase(TestCase):\n \"\"\"Edge case tests for duplicate detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_empty_photo_library(self):\n \"\"\"Test detection on empty library.\"\"\"\n count_exact = detect_exact_copies(self.user)\n count_visual = detect_visual_duplicates(self.user)\n \n self.assertEqual(count_exact, 0)\n self.assertEqual(count_visual, 0)\n\n def test_photo_without_files(self):\n \"\"\"Test photos without files are handled gracefully.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.image_hash = \"unique_hash\"\n photo.save()\n # No files attached\n \n count = detect_exact_copies(self.user)\n self.assertEqual(count, 0)\n\n def test_photo_with_short_file_hash(self):\n \"\"\"Test files with hash shorter than 32 chars.\"\"\"\n photo = create_test_photo(owner=self.user)\n file = File.objects.create(\n hash=\"short\", # Less than 32 chars\n path=\"/photos/test.jpg\",\n type=File.IMAGE,\n )\n photo.files.add(file)\n photo.save()\n \n # Should not raise\n count = detect_exact_copies(self.user)\n self.assertEqual(count, 0)\n\n def test_different_users_isolated(self):\n \"\"\"Test duplicates are isolated per user.\"\"\"\n other_user = create_test_user()\n \n # Create \"duplicate\" across users\n photo1 = create_test_photo(owner=self.user)\n photo1.image_hash = \"same_hash\"\n photo1.save()\n \n photo2 = create_test_photo(owner=other_user)\n photo2.image_hash = \"same_hash\"\n photo2.save()\n \n count = detect_exact_copies(self.user)\n \n # Should not find cross-user duplicates\n self.assertEqual(count, 0)\n\n def test_metadata_files_excluded(self):\n \"\"\"Test metadata files excluded from duplicate detection.\"\"\"\n photo = create_test_photo(owner=self.user)\n \n # Add metadata file\n metadata_file = File.objects.create(\n hash=\"a\" * 32 + \"suffix\",\n path=\"/photos/test.xmp\",\n type=File.METADATA_FILE,\n )\n photo.files.add(metadata_file)\n photo.save()\n \n # Should not count metadata file for duplicate detection\n count = detect_exact_copies(self.user)\n self.assertEqual(count, 0)\n\n def test_bktree_with_empty_hash(self):\n \"\"\"Test BKTree handles empty/None hash gracefully.\"\"\"\n def hamming(a, b):\n if not a or not b:\n return float('inf')\n return sum(c1 != c2 for c1, c2 in zip(a, b))\n \n tree = BKTree(hamming)\n tree.add(\"id1\", \"abc\")\n \n # Search with valid hash should work\n results = tree.search(\"abc\", 1)\n self.assertEqual(len(results), 1)\n\n def test_union_find_with_same_element_union(self):\n \"\"\"Test UnionFind handles self-union.\"\"\"\n uf = UnionFind()\n \n uf.union(\"a\", \"a\") # Self-union\n \n self.assertEqual(uf.find(\"a\"), \"a\")\n groups = uf.get_groups()\n self.assertEqual(len(groups), 0) # Single element not in groups\n\n def test_three_way_duplicate(self):\n \"\"\"Test detection with 3+ copies of same file.\"\"\"\n for _ in range(5):\n photo = create_test_photo(owner=self.user)\n photo.image_hash = \"five_way_duplicate\"\n photo.save()\n \n count = detect_exact_copies(self.user)\n \n self.assertEqual(count, 1)\n \n # Check all 5 photos in same group\n duplicate = Duplicate.objects.filter(owner=self.user).first()\n self.assertEqual(duplicate.photos.count(), 5)\n\n def test_transitive_duplicates_merged(self):\n \"\"\"Test transitive duplicates are merged into one group.\"\"\"\n # Photo A matches Photo B (same image_hash)\n # Photo B matches Photo C (same image_hash)\n # All three should be in same duplicate group\n \n photo_a = create_test_photo(owner=self.user)\n photo_a.image_hash = \"hash_abc\"\n file_a = File.objects.create(hash=\"x\" * 32, path=\"/a.jpg\", type=File.IMAGE)\n photo_a.files.add(file_a)\n photo_a.save()\n \n photo_b = create_test_photo(owner=self.user)\n photo_b.image_hash = \"hash_abc\" # Same as A\n file_b = File.objects.create(hash=\"y\" * 32, path=\"/b.jpg\", type=File.IMAGE)\n photo_b.files.add(file_b)\n photo_b.save()\n \n photo_c = create_test_photo(owner=self.user)\n photo_c.image_hash = \"hash_abc\" # Same as A and B\n file_c = File.objects.create(hash=\"z\" * 32, path=\"/c.jpg\", type=File.IMAGE)\n photo_c.files.add(file_c)\n photo_c.save()\n \n count = detect_exact_copies(self.user)\n \n # All 3 have same image_hash => all in one group\n self.assertEqual(count, 1)\n \n duplicate = Duplicate.objects.filter(owner=self.user).first()\n self.assertEqual(duplicate.photos.count(), 3)\n" }, { "path": "api/tests/test_duplicate_filtering.py", "content": "\"\"\"\nTests for Duplicate API filtering and detection edge cases.\n\nTests cover:\n- Filter by duplicate type (exact_copy, visual_duplicate)\n- Filter by status (pending, resolved, dismissed)\n- Photos without perceptual hash\n- Visual threshold sensitivity\n- Detection job handling\n\"\"\"\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.models.duplicate import Duplicate\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass DuplicateFilterByTypeTestCase(TestCase):\n \"\"\"Tests for filtering duplicates by type.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n \n # Create photos for duplicates\n self.photos = [create_test_photo(owner=self.user) for _ in range(6)]\n \n # Create exact copy duplicate\n self.exact_dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n self.exact_dup.photos.add(self.photos[0], self.photos[1])\n \n # Create visual duplicate\n self.visual_dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n similarity_score=0.95,\n )\n self.visual_dup.photos.add(self.photos[2], self.photos[3])\n \n # Create another exact copy\n self.exact_dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n self.exact_dup2.photos.add(self.photos[4], self.photos[5])\n\n def test_filter_exact_copies(self):\n \"\"\"Test filtering for exact copies only.\"\"\"\n response = self.client.get(\n f\"/api/duplicates?duplicate_type={Duplicate.DuplicateType.EXACT_COPY}\"\n )\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 2)\n for dup in response.data[\"results\"]:\n self.assertEqual(dup[\"duplicate_type\"], Duplicate.DuplicateType.EXACT_COPY)\n\n def test_filter_visual_duplicates(self):\n \"\"\"Test filtering for visual duplicates only.\"\"\"\n response = self.client.get(\n f\"/api/duplicates?duplicate_type={Duplicate.DuplicateType.VISUAL_DUPLICATE}\"\n )\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 1)\n self.assertEqual(\n response.data[\"results\"][0][\"duplicate_type\"],\n Duplicate.DuplicateType.VISUAL_DUPLICATE\n )\n\n def test_filter_invalid_type(self):\n \"\"\"Test filtering with invalid type returns empty or all.\"\"\"\n response = self.client.get(\"/api/duplicates?duplicate_type=invalid_type\")\n \n self.assertEqual(response.status_code, 200)\n # Should return all or empty depending on implementation\n self.assertIn(response.data[\"count\"], [0, 3])\n\n def test_no_filter_returns_all(self):\n \"\"\"Test that no filter returns all duplicates.\"\"\"\n response = self.client.get(\"/api/duplicates\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 3)\n\n\nclass DuplicateFilterByStatusTestCase(TestCase):\n \"\"\"Tests for filtering duplicates by status.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n \n # Create photos for duplicates\n self.photos = [create_test_photo(owner=self.user) for _ in range(6)]\n \n # Create pending duplicate\n self.pending = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n self.pending.photos.add(self.photos[0], self.photos[1])\n \n # Create resolved duplicate\n self.resolved = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n )\n self.resolved.photos.add(self.photos[2], self.photos[3])\n \n # Create dismissed duplicate\n self.dismissed = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n review_status=Duplicate.ReviewStatus.DISMISSED,\n )\n self.dismissed.photos.add(self.photos[4], self.photos[5])\n\n def test_filter_pending(self):\n \"\"\"Test filtering for pending duplicates.\"\"\"\n response = self.client.get(\n f\"/api/duplicates?status={Duplicate.ReviewStatus.PENDING}\"\n )\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 1)\n self.assertEqual(\n response.data[\"results\"][0][\"review_status\"],\n Duplicate.ReviewStatus.PENDING\n )\n\n def test_filter_resolved(self):\n \"\"\"Test filtering for resolved duplicates.\"\"\"\n response = self.client.get(\n f\"/api/duplicates?status={Duplicate.ReviewStatus.RESOLVED}\"\n )\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 1)\n\n def test_filter_dismissed(self):\n \"\"\"Test filtering for dismissed duplicates.\"\"\"\n response = self.client.get(\n f\"/api/duplicates?status={Duplicate.ReviewStatus.DISMISSED}\"\n )\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 1)\n\n def test_combined_type_and_status_filter(self):\n \"\"\"Test combining type and status filters.\"\"\"\n response = self.client.get(\n f\"/api/duplicates?duplicate_type={Duplicate.DuplicateType.EXACT_COPY}&status={Duplicate.ReviewStatus.PENDING}\"\n )\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 1)\n\n\nclass PhotosWithoutPerceptualHashTestCase(TestCase):\n \"\"\"Tests for handling photos without perceptual hash during detection.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_detection_handles_null_perceptual_hash(self):\n \"\"\"Test that detection handles photos with null perceptual hash.\"\"\"\n # Create photos, some without perceptual hash\n photo1 = create_test_photo(owner=self.user)\n _photo2 = create_test_photo(owner=self.user)\n \n # Set null perceptual hash\n photo1.image_phash = None\n photo1.save()\n \n # Detection should not crash - just trigger the endpoint\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_exact_copies\": True, \"detect_visual_duplicates\": True},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_visual_duplicate_detection_endpoint(self):\n \"\"\"Test visual duplicate detection API endpoint.\"\"\"\n # Create photos\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n # Set phash values\n photo1.image_phash = \"abcd1234\"\n photo1.save()\n photo2.image_phash = \"abcd1234\" # Same as photo1\n photo2.save()\n \n # Detection should work\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_visual_duplicates\": True, \"visual_threshold\": 5},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n\nclass DuplicateDetectionJobTestCase(TestCase):\n \"\"\"Tests for duplicate detection job handling.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_detect_with_all_options(self):\n \"\"\"Test detection with all options enabled.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\n \"detect_exact_copies\": True,\n \"detect_visual_duplicates\": True,\n \"visual_threshold\": 10,\n \"clear_pending\": False,\n },\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_detect_exact_only(self):\n \"\"\"Test detection with only exact copies.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_exact_copies\": True, \"detect_visual_duplicates\": False},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_detect_visual_only(self):\n \"\"\"Test detection with only visual duplicates.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_exact_copies\": False, \"detect_visual_duplicates\": True},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_detect_nothing_returns_error(self):\n \"\"\"Test detection with both options false.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_exact_copies\": False, \"detect_visual_duplicates\": False},\n format='json',\n )\n \n # Should return 400 or just skip detection\n self.assertIn(response.status_code, [200, 202, 400])\n\n def test_detect_with_clear_pending(self):\n \"\"\"Test detection with clear_pending option.\"\"\"\n # Create a pending duplicate first\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n dup.photos.add(photo1, photo2)\n \n # Detection API queues a background job\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_exact_copies\": True, \"clear_pending\": True},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n\nclass VisualThresholdTestCase(TestCase):\n \"\"\"Tests for visual duplicate threshold sensitivity.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_strict_threshold(self):\n \"\"\"Test visual detection with strict threshold (low value).\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_visual_duplicates\": True, \"visual_threshold\": 3},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_loose_threshold(self):\n \"\"\"Test visual detection with loose threshold (high value).\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_visual_duplicates\": True, \"visual_threshold\": 20},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_zero_threshold(self):\n \"\"\"Test visual detection with zero threshold (exact match only).\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_visual_duplicates\": True, \"visual_threshold\": 0},\n format='json',\n )\n \n self.assertIn(response.status_code, [200, 202])\n\n def test_negative_threshold_handled(self):\n \"\"\"Test that negative threshold is handled gracefully.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"detect_visual_duplicates\": True, \"visual_threshold\": -5},\n format='json',\n )\n \n # Should handle gracefully - either 400 or clamp to 0\n self.assertIn(response.status_code, [200, 202, 400])\n\n\nclass DuplicateListSortingTestCase(TestCase):\n \"\"\"Tests for duplicate list sorting.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n \n # Create multiple duplicates\n self.photos = [create_test_photo(owner=self.user) for _ in range(4)]\n \n self.dup1 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n self.dup1.photos.add(self.photos[0], self.photos[1])\n \n self.dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n similarity_score=0.95,\n )\n self.dup2.photos.add(self.photos[2], self.photos[3])\n\n def test_default_sorting(self):\n \"\"\"Test default sorting order.\"\"\"\n response = self.client.get(\"/api/duplicates\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"count\"], 2)\n\n def test_sort_by_created_at(self):\n \"\"\"Test sorting by created_at.\"\"\"\n response = self.client.get(\"/api/duplicates?ordering=-created_at\")\n \n self.assertEqual(response.status_code, 200)\n\n\nclass DuplicateBulkActionsTestCase(TestCase):\n \"\"\"Tests for bulk actions on duplicates.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_bulk_dismiss(self):\n \"\"\"Test bulk dismissing duplicates.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(4)]\n \n dup1 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup1.photos.add(photos[0], photos[1])\n \n dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup2.photos.add(photos[2], photos[3])\n \n # Dismiss first one\n response = self.client.post(f\"/api/duplicates/{dup1.id}/dismiss\")\n self.assertEqual(response.status_code, 200)\n \n dup1.refresh_from_db()\n self.assertEqual(dup1.review_status, Duplicate.ReviewStatus.DISMISSED)\n\n\nclass EmptyDuplicateGroupTestCase(TestCase):\n \"\"\"Tests for handling empty or invalid duplicate groups.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_duplicate_with_no_photos(self):\n \"\"\"Test handling duplicate group with no photos.\"\"\"\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n # Don't add any photos\n \n response = self.client.get(f\"/api/duplicates/{dup.id}\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"photo_count\"], 0)\n\n def test_duplicate_with_one_photo(self):\n \"\"\"Test handling duplicate group with only one photo.\"\"\"\n photo = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo)\n \n response = self.client.get(f\"/api/duplicates/{dup.id}\")\n \n self.assertEqual(response.status_code, 200)\n self.assertEqual(response.data[\"photo_count\"], 1)\n\n def test_resolve_single_photo_duplicate(self):\n \"\"\"Test resolving duplicate with only one photo.\"\"\"\n photo = create_test_photo(owner=self.user)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(photo)\n \n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n {\"keep_photo_hash\": photo.image_hash},\n format='json',\n )\n \n # Should succeed but with no photos to trash\n self.assertIn(response.status_code, [200, 400])\n" }, { "path": "api/tests/test_edge_cases_integration.py", "content": "\"\"\"\nTests for edge cases and error handling.\n\nTests:\n- Photos with no EXIF data\n- Missing/corrupted files\n- Concurrent detection jobs\n- Empty/invalid data handling\n\"\"\"\n\n\nfrom django.test import TestCase, TransactionTestCase\nfrom rest_framework.test import APIClient, APITestCase\n\nfrom api.models.duplicate import Duplicate\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.photo_metadata import PhotoMetadata\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass PhotoNoExifDataTestCase(TestCase):\n \"\"\"Test handling of photos with no EXIF data.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_photo_without_exif_timestamp(self):\n \"\"\"Test handling photo with no EXIF timestamp.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = None\n photo.save()\n \n # Should still be usable\n self.assertIsNotNone(photo.pk)\n self.assertIsNone(photo.exif_timestamp)\n\n def test_photo_without_gps_data(self):\n \"\"\"Test handling photo with no GPS data.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.exif_gps_lat = None\n photo.exif_gps_lon = None\n photo.save()\n \n # Should still be usable\n self.assertIsNotNone(photo.pk)\n\n def test_photo_without_perceptual_hash(self):\n \"\"\"Test handling photo with no perceptual hash.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.image_phash = None\n photo.save()\n \n # Should still be usable but not in visual duplicate detection\n self.assertIsNotNone(photo.pk)\n self.assertIsNone(photo.image_phash)\n\n def test_stack_with_no_exif_photos(self):\n \"\"\"Test creating stack with photos that have no EXIF.\"\"\"\n photos = []\n for _ in range(3):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = None\n photo.save()\n photos.append(photo)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos)\n \n # auto_select_primary should still work\n _result = stack.auto_select_primary()\n # May or may not select one depending on implementation\n # The important thing is it doesn't crash\n\n def test_duplicate_with_no_metadata_photos(self):\n \"\"\"Test duplicate group with photos lacking metadata.\"\"\"\n photos = []\n for _ in range(2):\n photo = create_test_photo(owner=self.user)\n # Don't create metadata\n PhotoMetadata.objects.filter(photo=photo).delete()\n photos.append(photo)\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup.photos.add(*photos)\n \n # auto_select_best_photo should handle gracefully\n _result = dup.auto_select_best_photo()\n # Should return something (or None) without crashing\n\n def test_burst_detection_no_timestamps(self):\n \"\"\"Test burst detection with photos lacking timestamps.\"\"\"\n photos = []\n for _ in range(5):\n photo = create_test_photo(owner=self.user)\n photo.exif_timestamp = None\n photo.save()\n photos.append(photo)\n \n # Should not create burst stacks for photos without timestamps\n # (timestamps are required for burst proximity detection)\n\n\nclass MissingFileTestCase(TestCase):\n \"\"\"Test handling of photos with missing files.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_photo_with_null_main_file(self):\n \"\"\"Test handling photo with null main_file reference.\"\"\"\n photo = create_test_photo(owner=self.user)\n \n # This might not be allowed by the model, but test graceful handling\n # Note: Can't actually set main_file to None due to NOT NULL constraint\n # So we test that the photo with a valid file still works\n self.assertIsNotNone(photo.main_file)\n\n def test_stack_photos_with_missing_metadata(self):\n \"\"\"Test stack with photos that have no PhotoMetadata records.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(3)]\n \n # Delete metadata records\n PhotoMetadata.objects.filter(photo__in=photos).delete()\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos)\n \n # Stack should still function\n self.assertEqual(stack.photos.count(), 3)\n\n\nclass ConcurrentDetectionTestCase(TransactionTestCase):\n \"\"\"Test concurrent detection job handling.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_concurrent_duplicate_detection_requests(self):\n \"\"\"Test handling multiple simultaneous detection requests.\"\"\"\n # Create photos\n for _ in range(5):\n create_test_photo(owner=self.user)\n \n results = []\n errors = []\n \n def trigger_detection():\n try:\n response = self.client.post(\"/api/duplicates/detect\")\n results.append(response.status_code)\n except Exception as e:\n errors.append(str(e))\n \n # Trigger multiple detections (simulated - they run sequentially in test)\n for _ in range(3):\n trigger_detection()\n \n # All requests should succeed (or be queued)\n for status in results:\n self.assertIn(status, [200, 202, 409]) # 409 = conflict if already running\n \n # No errors should occur\n self.assertEqual(len(errors), 0)\n\n def test_concurrent_stack_detection_requests(self):\n \"\"\"Test handling multiple simultaneous stack detection requests.\"\"\"\n # Create photos\n for _ in range(5):\n create_test_photo(owner=self.user)\n \n results = []\n \n for _ in range(3):\n response = self.client.post(\"/api/stacks/detect\")\n results.append(response.status_code)\n \n # All requests should succeed or be handled gracefully\n for status in results:\n self.assertIn(status, [200, 202, 409])\n\n\nclass EmptyDataTestCase(APITestCase):\n \"\"\"Test handling of empty or minimal data.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_duplicate_list_empty(self):\n \"\"\"Test duplicate list with no duplicates.\"\"\"\n response = self.client.get(\"/api/duplicates\")\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data.get(\"results\", [])), 0)\n\n def test_stack_list_empty(self):\n \"\"\"Test stack list with no stacks.\"\"\"\n response = self.client.get(\"/api/stacks\")\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data.get(\"results\", [])), 0)\n\n def test_duplicate_stats_empty(self):\n \"\"\"Test duplicate stats with no data.\"\"\"\n response = self.client.get(\"/api/duplicates/stats\")\n self.assertEqual(response.status_code, 200)\n\n def test_stack_stats_empty(self):\n \"\"\"Test stack stats with no data.\"\"\"\n response = self.client.get(\"/api/stacks/stats\")\n self.assertEqual(response.status_code, 200)\n\n def test_detection_with_no_photos(self):\n \"\"\"Test detection when user has no photos.\"\"\"\n response = self.client.post(\"/api/duplicates/detect\")\n # Should succeed but find nothing\n self.assertIn(response.status_code, [200, 202])\n\n def test_stack_detection_with_no_photos(self):\n \"\"\"Test stack detection when user has no photos.\"\"\"\n response = self.client.post(\"/api/stacks/detect\")\n # Should succeed but find nothing\n self.assertIn(response.status_code, [200, 202])\n\n\nclass InvalidDataTestCase(APITestCase):\n \"\"\"Test handling of invalid data inputs.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n\n def test_resolve_with_invalid_photo_id(self):\n \"\"\"Test resolve with invalid photo ID.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n {\"kept_photo_id\": \"not-a-valid-uuid\"},\n format=\"json\"\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_add_to_stack_with_invalid_photo_ids(self):\n \"\"\"Test adding invalid photo IDs to stack.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(2)]\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos)\n \n response = self.client.post(\n f\"/api/stacks/{stack.id}/add\",\n {\"photo_ids\": [\"invalid-id-1\", \"invalid-id-2\"]},\n format=\"json\"\n )\n # Should handle gracefully\n self.assertIn(response.status_code, [200, 400, 404])\n\n def test_set_primary_with_invalid_photo_id(self):\n \"\"\"Test setting primary with invalid photo ID.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(2)]\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos)\n \n response = self.client.post(\n f\"/api/stacks/{stack.id}/primary\",\n {\"photo_id\": \"not-a-uuid\"},\n format=\"json\"\n )\n self.assertIn(response.status_code, [400, 404])\n\n def test_detection_with_invalid_options(self):\n \"\"\"Test detection with invalid options.\"\"\"\n response = self.client.post(\n \"/api/duplicates/detect\",\n {\"invalid_option\": \"value\"},\n format=\"json\"\n )\n # Should ignore invalid options and proceed\n self.assertIn(response.status_code, [200, 202, 400])\n\n\nclass SinglePhotoGroupTestCase(TestCase):\n \"\"\"Test handling of single-photo groups.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_duplicate_with_single_photo_deleted(self):\n \"\"\"Test duplicate group cleanup when reduced to single photo.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n # Remove one photo\n dup.photos.remove(photos[0])\n \n # Group should still exist but may be cleaned up depending on implementation\n # The important thing is no crash occurs\n\n def test_stack_with_single_photo(self):\n \"\"\"Test stack behavior with single photo.\"\"\"\n photo = create_test_photo(owner=self.user)\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(photo)\n \n # Single photo stack is valid for manual stacks\n self.assertEqual(stack.photos.count(), 1)\n\n def test_auto_select_with_single_photo(self):\n \"\"\"Test auto_select_primary with single photo.\"\"\"\n photo = create_test_photo(owner=self.user)\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(photo)\n \n _result = stack.auto_select_primary()\n stack.refresh_from_db()\n \n # Should select the only photo\n self.assertEqual(stack.primary_photo, photo)\n\n\nclass PhotoDeletionEdgeCasesTestCase(TestCase):\n \"\"\"Test edge cases around photo deletion.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_delete_photo_in_multiple_stacks(self):\n \"\"\"Test deleting a photo that's in multiple stacks.\"\"\"\n photo = create_test_photo(owner=self.user)\n other_photos1 = [create_test_photo(owner=self.user) for _ in range(2)]\n other_photos2 = [create_test_photo(owner=self.user) for _ in range(2)]\n \n stack1 = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack1.photos.add(photo, *other_photos1)\n \n stack2 = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack2.photos.add(photo, *other_photos2)\n \n # Delete the shared photo\n photo.manual_delete()\n \n # Both stacks should still exist with remaining photos\n stack1.refresh_from_db()\n stack2.refresh_from_db()\n self.assertEqual(stack1.photos.count(), 2)\n self.assertEqual(stack2.photos.count(), 2)\n\n def test_delete_photo_in_multiple_duplicate_groups(self):\n \"\"\"Test deleting a photo that's in multiple duplicate groups.\"\"\"\n photo = create_test_photo(owner=self.user)\n other_photos1 = [create_test_photo(owner=self.user)]\n other_photos2 = [create_test_photo(owner=self.user)]\n \n dup1 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup1.photos.add(photo, *other_photos1)\n \n dup2 = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.VISUAL_DUPLICATE,\n )\n dup2.photos.add(photo, *other_photos2)\n \n # Delete the shared photo\n photo.manual_delete()\n \n # Both groups should be cleaned up (single photo remaining)\n # Depending on implementation, they may be deleted or left with 1 photo\n\n def test_delete_primary_photo_from_stack(self):\n \"\"\"Test deleting the primary photo from a stack.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(3)]\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n primary_photo=photos[0],\n )\n stack.photos.add(*photos)\n \n # Delete the primary photo\n photos[0].manual_delete()\n \n stack.refresh_from_db()\n \n # Stack should still exist\n self.assertEqual(stack.photos.count(), 2)\n # Primary may be cleared or set to another photo depending on implementation\n\n\nclass MetadataEdgeCasesTestCase(TestCase):\n \"\"\"Test edge cases for metadata handling.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_photo_with_extreme_dimensions(self):\n \"\"\"Test handling photos with extreme dimensions.\"\"\"\n photo = create_test_photo(owner=self.user)\n metadata, _ = PhotoMetadata.objects.get_or_create(photo=photo)\n \n # Very large dimensions\n metadata.width = 50000\n metadata.height = 50000\n metadata.save()\n \n # Should not crash on resolution calculation\n self.assertIsNotNone(metadata.resolution)\n\n def test_photo_with_zero_dimensions(self):\n \"\"\"Test handling photos with zero dimensions.\"\"\"\n photo = create_test_photo(owner=self.user)\n metadata, _ = PhotoMetadata.objects.get_or_create(photo=photo)\n \n metadata.width = 0\n metadata.height = 0\n metadata.save()\n \n # Should handle gracefully\n self.assertEqual(metadata.width, 0)\n\n def test_duplicate_savings_with_zero_size(self):\n \"\"\"Test potential savings calculation with zero-size photos.\"\"\"\n photos = [create_test_photo(owner=self.user) for _ in range(3)]\n for photo in photos:\n photo.size = 0\n photo.save()\n \n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n savings = dup.calculate_potential_savings()\n self.assertEqual(savings, 0)\n" }, { "path": "api/tests/test_edit_photo_details.py", "content": "from unittest.mock import patch\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass EditPhotoDetailsTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n @patch(\"api.models.Photo._extract_date_time_from_exif\", autospec=True)\n def test_should_update_timestamp(self, extract_date_time_from_exif_mock):\n photo = create_test_photo(owner=self.user1)\n\n payload = {\"exif_timestamp\": \"1970-01-01T00:00:00.001Z\"}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.patch(\n f\"/api/photos/edit/{photo.image_hash}/\",\n format=\"json\",\n data=payload,\n headers=headers,\n )\n data = response.json()\n\n self.assertEqual(200, response.status_code)\n self.assertEqual(\"1970-01-01T00:00:00.001000Z\", data[\"timestamp\"])\n self.assertEqual(photo.image_hash, data[\"image_hash\"])\n self.assertIsNone(data[\"exif_timestamp\"])\n self.assertEqual(0, data[\"rating\"])\n self.assertFalse(data[\"hidden\"])\n self.assertFalse(data[\"in_trashcan\"])\n self.assertFalse(data[\"video\"])\n extract_date_time_from_exif_mock.assert_called()\n\n @patch(\"api.models.Photo._extract_date_time_from_exif\", autospec=True)\n def test_should_not_update_other_properties(self, extract_date_time_from_exif_mock):\n photo = create_test_photo(owner=self.user1)\n\n payload = {\n \"timestamp\": \"1970-01-01T00:00:00.001Z\",\n \"image_hash\": \"BLAH-BLAH-BLAH-BLAH\",\n \"rating\": 100,\n \"deleted\": True,\n \"hidden\": True,\n \"in_trashcan\": True,\n \"video\": True,\n }\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.patch(\n f\"/api/photos/edit/{photo.image_hash}/\",\n format=\"json\",\n data=payload,\n headers=headers,\n )\n data = response.json()\n\n self.assertEqual(200, response.status_code)\n self.assertNotEqual(payload[\"timestamp\"], data[\"timestamp\"])\n self.assertNotEqual(payload[\"image_hash\"], data[\"image_hash\"])\n self.assertNotEqual(payload[\"rating\"], data[\"rating\"])\n self.assertNotEqual(payload[\"hidden\"], data[\"hidden\"])\n self.assertNotEqual(payload[\"in_trashcan\"], data[\"in_trashcan\"])\n self.assertNotEqual(payload[\"video\"], data[\"video\"])\n extract_date_time_from_exif_mock.assert_not_called()\n" }, { "path": "api/tests/test_face_extractor.py", "content": "\"\"\"Tests for face_extractor module.\"\"\"\n\nfrom unittest.mock import MagicMock, patch\n\nfrom django.test import TestCase\n\nfrom api import face_extractor\n\n\nclass FaceExtractorTest(TestCase):\n \"\"\"Test face extraction functionality.\"\"\"\n\n @patch(\"api.face_extractor.get_face_locations\")\n def test_extract_from_dlib_handles_exception(self, mock_get_face_locations):\n \"\"\"Test that extract_from_dlib returns empty list when exception occurs.\"\"\"\n # Setup: make get_face_locations raise an exception\n mock_get_face_locations.side_effect = Exception(\"Test exception\")\n\n # Create a mock owner with face_recognition_model\n mock_owner = MagicMock()\n mock_owner.face_recognition_model = \"hog\"\n\n # Call the function\n result = face_extractor.extract_from_dlib(\n image_path=\"/path/to/image.jpg\",\n big_thumbnail_path=\"/path/to/thumbnail.jpg\",\n owner=mock_owner,\n )\n\n # Verify that it returns an empty list instead of raising UnboundLocalError\n self.assertEqual(result, [])\n\n @patch(\"api.face_extractor.get_face_locations\")\n def test_extract_from_dlib_success(self, mock_get_face_locations):\n \"\"\"Test that extract_from_dlib works correctly on success.\"\"\"\n # Setup: make get_face_locations return some face locations\n mock_face_locations = [(10, 20, 30, 40), (50, 60, 70, 80)]\n mock_get_face_locations.return_value = mock_face_locations\n\n # Create a mock owner with face_recognition_model\n mock_owner = MagicMock()\n mock_owner.face_recognition_model = \"hog\"\n\n # Call the function\n result = face_extractor.extract_from_dlib(\n image_path=\"/path/to/image.jpg\",\n big_thumbnail_path=\"/path/to/thumbnail.jpg\",\n owner=mock_owner,\n )\n\n # Verify that it returns face locations with None appended\n expected = [(10, 20, 30, 40, None), (50, 60, 70, 80, None)]\n self.assertEqual(result, expected)\n\n @patch(\"api.face_extractor.extract_from_exif\")\n @patch(\"api.face_extractor.extract_from_dlib\")\n def test_extract_prefers_exif(self, mock_dlib, mock_exif):\n \"\"\"Test that extract function prefers EXIF data over dlib.\"\"\"\n # Setup: make extract_from_exif return some data\n mock_exif_data = [(10, 20, 30, 40, \"John Doe\")]\n mock_exif.return_value = mock_exif_data\n\n mock_owner = MagicMock()\n\n # Call the function\n result = face_extractor.extract(\n image_path=\"/path/to/image.jpg\",\n big_thumbnail_path=\"/path/to/thumbnail.jpg\",\n owner=mock_owner,\n )\n\n # Verify that it returns EXIF data and doesn't call dlib\n self.assertEqual(result, mock_exif_data)\n mock_dlib.assert_not_called()\n\n @patch(\"api.face_extractor.extract_from_exif\")\n @patch(\"api.face_extractor.extract_from_dlib\")\n def test_extract_fallback_to_dlib(self, mock_dlib, mock_exif):\n \"\"\"Test that extract function falls back to dlib when no EXIF data.\"\"\"\n # Setup: make extract_from_exif return None\n mock_exif.return_value = None\n mock_dlib_data = [(10, 20, 30, 40, None)]\n mock_dlib.return_value = mock_dlib_data\n\n mock_owner = MagicMock()\n\n # Call the function\n result = face_extractor.extract(\n image_path=\"/path/to/image.jpg\",\n big_thumbnail_path=\"/path/to/thumbnail.jpg\",\n owner=mock_owner,\n )\n\n # Verify that it returns dlib data\n self.assertEqual(result, mock_dlib_data)\n mock_dlib.assert_called_once()\n" }, { "path": "api/tests/test_face_writeback.py", "content": "from unittest.mock import MagicMock, patch\n\nfrom django.test import TestCase\n\nfrom api.metadata.face_regions import (\n build_face_region_exiftool_args,\n get_face_region_tags,\n reverse_orientation_transform,\n thumbnail_coords_to_normalized,\n)\nfrom api.models.person import Person\nfrom api.tests.utils import (\n create_test_face,\n create_test_person,\n create_test_photo,\n create_test_user,\n)\n\n\nclass TestThumbnailCoordsToNormalized(TestCase):\n def test_basic_conversion(self):\n \"\"\"Known pixel coords should produce expected normalized values.\"\"\"\n # Face at center of a 1000x800 thumbnail\n # top=300, right=600, bottom=500, left=400\n x, y, w, h = thumbnail_coords_to_normalized(\n top=300,\n right=600,\n bottom=500,\n left=400,\n thumb_width=1000,\n thumb_height=800,\n )\n self.assertAlmostEqual(x, 0.5) # center_x = (400+600)/2/1000\n self.assertAlmostEqual(y, 0.5) # center_y = (300+500)/2/800\n self.assertAlmostEqual(w, 0.2) # w = (600-400)/1000\n self.assertAlmostEqual(h, 0.25) # h = (500-300)/800\n\n def test_corner_face(self):\n \"\"\"Face in top-left corner.\"\"\"\n x, y, w, h = thumbnail_coords_to_normalized(\n top=0,\n right=100,\n bottom=100,\n left=0,\n thumb_width=1000,\n thumb_height=1000,\n )\n self.assertAlmostEqual(x, 0.05)\n self.assertAlmostEqual(y, 0.05)\n self.assertAlmostEqual(w, 0.1)\n self.assertAlmostEqual(h, 0.1)\n\n\nclass TestReverseOrientationTransform(TestCase):\n def test_identity_for_normal_orientation(self):\n \"\"\"Normal orientation should be a no-op.\"\"\"\n x, y, w, h = reverse_orientation_transform(\n 0.5, 0.3, 0.2, 0.1, \"Horizontal (normal)\"\n )\n self.assertAlmostEqual(x, 0.5)\n self.assertAlmostEqual(y, 0.3)\n self.assertAlmostEqual(w, 0.2)\n self.assertAlmostEqual(h, 0.1)\n\n def test_identity_for_none_orientation(self):\n \"\"\"None orientation should be a no-op.\"\"\"\n x, y, w, h = reverse_orientation_transform(0.5, 0.3, 0.2, 0.1, None)\n self.assertAlmostEqual(x, 0.5)\n self.assertAlmostEqual(y, 0.3)\n self.assertAlmostEqual(w, 0.2)\n self.assertAlmostEqual(h, 0.1)\n\n def test_round_trip_rotate_90_cw(self):\n \"\"\"Forward then reverse for Rotate 90 CW should return original coords.\"\"\"\n self._test_round_trip(\"Rotate 90 CW\")\n\n def test_round_trip_mirror_horizontal(self):\n self._test_round_trip(\"Mirror horizontal\")\n\n def test_round_trip_rotate_180(self):\n self._test_round_trip(\"Rotate 180\")\n\n def test_round_trip_mirror_vertical(self):\n self._test_round_trip(\"Mirror vertical\")\n\n def test_round_trip_rotate_270_cw(self):\n self._test_round_trip(\"Rotate 270 CW\")\n\n def test_round_trip_mirror_horizontal_rotate_90_cw(self):\n self._test_round_trip(\"Mirror horizontal and rotate 90 CW\")\n\n def _test_round_trip(self, orientation):\n \"\"\"Apply forward transform (from face_extractor) then reverse, verify identity.\"\"\"\n orig_x, orig_y, orig_w, orig_h = 0.4, 0.3, 0.2, 0.15\n\n # Apply forward transform (same logic as face_extractor.py lines 54-80)\n correct_x, correct_y = orig_x, orig_y\n correct_w, correct_h = orig_w, orig_h\n if orientation == \"Rotate 90 CW\":\n temp_x = correct_x\n correct_x = 1 - correct_y\n correct_y = temp_x\n correct_w, correct_h = correct_h, correct_w\n elif orientation == \"Mirror horizontal\":\n correct_x = 1 - correct_x\n elif orientation == \"Rotate 180\":\n correct_x = 1 - correct_x\n correct_y = 1 - correct_y\n elif orientation == \"Mirror vertical\":\n correct_y = 1 - correct_y\n elif orientation == \"Mirror horizontal and rotate 270 CW\":\n temp_x = correct_x\n correct_x = 1 - correct_y\n correct_y = temp_x\n correct_w, correct_h = correct_h, correct_w\n elif orientation == \"Mirror horizontal and rotate 90 CW\":\n temp_x = correct_x\n correct_x = correct_y\n correct_y = 1 - temp_x\n correct_w, correct_h = correct_h, correct_w\n elif orientation == \"Rotate 270 CW\":\n temp_x = correct_x\n correct_x = correct_y\n correct_y = 1 - temp_x\n correct_w, correct_h = correct_h, correct_w\n\n # Now reverse\n rx, ry, rw, rh = reverse_orientation_transform(\n correct_x, correct_y, correct_w, correct_h, orientation\n )\n self.assertAlmostEqual(\n rx, orig_x, places=10, msg=f\"x mismatch for {orientation}\"\n )\n self.assertAlmostEqual(\n ry, orig_y, places=10, msg=f\"y mismatch for {orientation}\"\n )\n self.assertAlmostEqual(\n rw, orig_w, places=10, msg=f\"w mismatch for {orientation}\"\n )\n self.assertAlmostEqual(\n rh, orig_h, places=10, msg=f\"h mismatch for {orientation}\"\n )\n\n\nclass TestBuildFaceRegionExiftoolArgs(TestCase):\n def test_single_face(self):\n \"\"\"Single face region should produce correct structured tag.\"\"\"\n regions = [{\"name\": \"Alice\", \"x\": 0.5, \"y\": 0.3, \"w\": 0.2, \"h\": 0.15}]\n result = build_face_region_exiftool_args(regions)\n self.assertIn(\"XMP-mwg-rs:RegionInfo\", result)\n value = result[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Alice\", value)\n self.assertIn(\"Type=Face\", value)\n self.assertIn(\"Unit=normalized\", value)\n self.assertIn(\"RegionList=\", value)\n\n def test_multiple_faces(self):\n \"\"\"Multiple face regions should all appear in RegionList.\"\"\"\n regions = [\n {\"name\": \"Alice\", \"x\": 0.3, \"y\": 0.3, \"w\": 0.1, \"h\": 0.1},\n {\"name\": \"Bob\", \"x\": 0.7, \"y\": 0.5, \"w\": 0.15, \"h\": 0.2},\n {\"name\": \"Charlie\", \"x\": 0.5, \"y\": 0.8, \"w\": 0.12, \"h\": 0.1},\n ]\n result = build_face_region_exiftool_args(regions)\n value = result[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Alice\", value)\n self.assertIn(\"Bob\", value)\n self.assertIn(\"Charlie\", value)\n\n def test_special_characters_in_name(self):\n \"\"\"Person names with commas, braces, equals should be escaped.\"\"\"\n regions = [{\"name\": \"O'Brien, Jr.\", \"x\": 0.5, \"y\": 0.5, \"w\": 0.1, \"h\": 0.1}]\n result = build_face_region_exiftool_args(regions)\n value = result[\"XMP-mwg-rs:RegionInfo\"]\n # Comma should be escaped\n self.assertIn(\"O'Brien\\\\, Jr.\", value)\n\n def test_escape_braces_and_equals(self):\n \"\"\"Braces and equals in names should be escaped.\"\"\"\n regions = [{\"name\": \"Test{=}\", \"x\": 0.5, \"y\": 0.5, \"w\": 0.1, \"h\": 0.1}]\n result = build_face_region_exiftool_args(regions)\n value = result[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Test\\\\{\\\\=\\\\}\", value)\n\n def test_applied_to_dimensions(self):\n \"\"\"AppliedToDimensions should be included when image dimensions are provided.\"\"\"\n regions = [{\"name\": \"Alice\", \"x\": 0.5, \"y\": 0.3, \"w\": 0.2, \"h\": 0.15}]\n result = build_face_region_exiftool_args(regions, image_width=4000, image_height=3000)\n value = result[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"AppliedToDimensions={W=4000,H=3000,Unit=pixel}\", value)\n\n def test_no_applied_to_dimensions_when_missing(self):\n \"\"\"AppliedToDimensions should be omitted when image dimensions are not available.\"\"\"\n regions = [{\"name\": \"Alice\", \"x\": 0.5, \"y\": 0.3, \"w\": 0.2, \"h\": 0.15}]\n result = build_face_region_exiftool_args(regions)\n value = result[\"XMP-mwg-rs:RegionInfo\"]\n self.assertNotIn(\"AppliedToDimensions\", value)\n\n def test_subject_keywords_for_named_faces(self):\n \"\"\"Named faces should be added as XMP:Subject keywords.\"\"\"\n regions = [\n {\"name\": \"Alice\", \"x\": 0.3, \"y\": 0.3, \"w\": 0.1, \"h\": 0.1},\n {\"name\": \"Bob\", \"x\": 0.7, \"y\": 0.5, \"w\": 0.15, \"h\": 0.2},\n {\"name\": \"\", \"x\": 0.5, \"y\": 0.8, \"w\": 0.12, \"h\": 0.1},\n ]\n result = build_face_region_exiftool_args(regions)\n self.assertIn(\"XMP:Subject\", result)\n self.assertEqual(result[\"XMP:Subject\"], [\"Alice\", \"Bob\"])\n\n def test_no_subject_keywords_when_all_unnamed(self):\n \"\"\"No XMP:Subject should be set when all faces are unnamed.\"\"\"\n regions = [\n {\"name\": \"\", \"x\": 0.3, \"y\": 0.3, \"w\": 0.1, \"h\": 0.1},\n ]\n result = build_face_region_exiftool_args(regions)\n self.assertNotIn(\"XMP:Subject\", result)\n\n\nclass TestRoundTripCoordinates(TestCase):\n def test_round_trip_no_orientation(self):\n \"\"\"Pixel coords -> normalize -> (simulate XMP read-back) -> verify ~= original.\"\"\"\n thumb_width = 1000\n thumb_height = 800\n orig_top, orig_right, orig_bottom, orig_left = 200, 600, 400, 400\n\n # Step 1: Convert pixel -> normalized (writeback path)\n x, y, w, h = thumbnail_coords_to_normalized(\n orig_top,\n orig_right,\n orig_bottom,\n orig_left,\n thumb_width,\n thumb_height,\n )\n\n # Step 2: Simulate read-back (face_extractor.py lines 82-90)\n half_width = (w * thumb_width) / 2\n half_height = (h * thumb_height) / 2\n read_top = int((y * thumb_height) - half_height)\n read_right = int((x * thumb_width) + half_width)\n read_bottom = int((y * thumb_height) + half_height)\n read_left = int((x * thumb_width) - half_width)\n\n # Verify within 1px tolerance (int rounding)\n self.assertAlmostEqual(read_top, orig_top, delta=1)\n self.assertAlmostEqual(read_right, orig_right, delta=1)\n self.assertAlmostEqual(read_bottom, orig_bottom, delta=1)\n self.assertAlmostEqual(read_left, orig_left, delta=1)\n\n\nclass TestGetFaceRegionTags(TestCase):\n def setUp(self):\n self.user = create_test_user()\n\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_returns_tags_for_labeled_faces(self, mock_pil_open, mock_get_metadata):\n \"\"\"get_face_region_tags should return a dict with RegionInfo for labeled faces.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n person = create_test_person(\n name=\"Alice\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n tags = get_face_region_tags(photo)\n\n self.assertIn(\"XMP-mwg-rs:RegionInfo\", tags)\n self.assertIn(\"Alice\", tags[\"XMP-mwg-rs:RegionInfo\"])\n # AppliedToDimensions should be present\n self.assertIn(\n \"AppliedToDimensions={W=4000,H=3000,Unit=pixel}\",\n tags[\"XMP-mwg-rs:RegionInfo\"],\n )\n # XMP:Subject should contain the person name\n self.assertIn(\"XMP:Subject\", tags)\n self.assertEqual(tags[\"XMP:Subject\"], [\"Alice\"])\n\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_returns_all_faces(self, mock_pil_open, mock_get_metadata):\n \"\"\"Photo with 3 labeled faces should have all 3 in the returned tags.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n for name in [\"Alice\", \"Bob\", \"Charlie\"]:\n person = create_test_person(\n name=name, kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n tags = get_face_region_tags(photo)\n\n value = tags[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Alice\", value)\n self.assertIn(\"Bob\", value)\n self.assertIn(\"Charlie\", value)\n\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_unlabeled_faces_written_with_empty_name(\n self, mock_pil_open, mock_get_metadata\n ):\n \"\"\"Faces without a KIND_USER person should be written with an empty name.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n cluster_person = create_test_person(\n name=\"cluster_0001\", kind=Person.KIND_CLUSTER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=cluster_person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n tags = get_face_region_tags(photo)\n\n self.assertIn(\"XMP-mwg-rs:RegionInfo\", tags)\n value = tags[\"XMP-mwg-rs:RegionInfo\"]\n # Should have the face region but with empty name\n self.assertIn(\"Name=,Type=Face\", value)\n self.assertNotIn(\"cluster_0001\", value)\n\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_faces_with_no_person_written_with_empty_name(\n self, mock_pil_open, mock_get_metadata\n ):\n \"\"\"Faces with person=None should be written with an empty name.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n create_test_face(\n photo=photo,\n person=None,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n tags = get_face_region_tags(photo)\n\n self.assertIn(\"XMP-mwg-rs:RegionInfo\", tags)\n value = tags[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Name=,Type=Face\", value)\n\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_mixed_labeled_and_unlabeled_faces(\n self, mock_pil_open, mock_get_metadata\n ):\n \"\"\"Photo with both labeled and unlabeled faces should include all.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n labeled_person = create_test_person(\n name=\"Alice\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=labeled_person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n cluster_person = create_test_person(\n name=\"cluster_0001\", kind=Person.KIND_CLUSTER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=cluster_person,\n location_top=400,\n location_right=600,\n location_bottom=600,\n location_left=400,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n tags = get_face_region_tags(photo)\n\n value = tags[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Alice\", value)\n self.assertNotIn(\"cluster_0001\", value)\n # Should have 2 face regions\n self.assertEqual(value.count(\"Type=Face\"), 2)\n\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_skips_deleted_faces(self, mock_pil_open, mock_get_metadata):\n \"\"\"Deleted faces should not be included in the tags.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n person = create_test_person(\n name=\"Active\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n deleted_person = create_test_person(\n name=\"Deleted\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=deleted_person,\n deleted=True,\n location_top=400,\n location_right=600,\n location_bottom=600,\n location_left=400,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n tags = get_face_region_tags(photo)\n\n value = tags[\"XMP-mwg-rs:RegionInfo\"]\n self.assertIn(\"Active\", value)\n self.assertNotIn(\"Deleted\", value)\n\n\nclass TestSaveMetadataIntegration(TestCase):\n def setUp(self):\n self.user = create_test_user()\n\n @patch(\"api.models.photo.write_metadata\")\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_save_metadata_with_face_tags(\n self, mock_pil_open, mock_get_metadata, mock_write_metadata\n ):\n \"\"\"_save_metadata(metadata_types=[\"face_tags\"]) should write face regions.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n person = create_test_person(\n name=\"Test Person\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n photo._save_metadata(use_sidecar=True, metadata_types=[\"face_tags\"])\n\n mock_write_metadata.assert_called_once()\n tags = mock_write_metadata.call_args[0][1]\n self.assertIn(\"XMP-mwg-rs:RegionInfo\", tags)\n self.assertIn(\"Test Person\", tags[\"XMP-mwg-rs:RegionInfo\"])\n\n @patch(\"api.models.photo.write_metadata\")\n def test_save_metadata_default_does_not_write_face_tags(self, mock_write_metadata):\n \"\"\"_save_metadata() with default args should NOT write face tags.\"\"\"\n photo = create_test_photo(owner=self.user)\n person = create_test_person(\n name=\"Test Person\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n # Default call (no metadata_types) — should only consider ratings\n photo._save_metadata()\n\n # Rating is 0 by default and there are no modified_fields=None,\n # so it will write the rating tag\n if mock_write_metadata.called:\n tags = mock_write_metadata.call_args[0][1]\n self.assertNotIn(\"XMP-mwg-rs:RegionInfo\", tags)\n\n @patch(\"api.models.photo.write_metadata\")\n @patch(\"api.metadata.face_regions.get_metadata\")\n @patch(\"api.metadata.face_regions.PIL.Image.open\")\n def test_save_metadata_combined_types(\n self, mock_pil_open, mock_get_metadata, mock_write_metadata\n ):\n \"\"\"_save_metadata with both types should write ratings AND face tags together.\"\"\"\n photo = create_test_photo(\n owner=self.user, thumbnail_big=\"thumbnails_big/test.jpg\"\n )\n photo.rating = 5\n person = create_test_person(\n name=\"Alice\", kind=Person.KIND_USER, cluster_owner=self.user\n )\n create_test_face(\n photo=photo,\n person=person,\n location_top=100,\n location_right=300,\n location_bottom=300,\n location_left=100,\n )\n\n mock_img = MagicMock()\n mock_img.size = (1000, 800)\n mock_pil_open.return_value = mock_img\n mock_get_metadata.return_value = (None, 4000, 3000)\n\n photo._save_metadata(use_sidecar=True, metadata_types=[\"ratings\", \"face_tags\"])\n\n mock_write_metadata.assert_called_once()\n tags = mock_write_metadata.call_args[0][1]\n self.assertIn(\"Rating\", tags)\n self.assertEqual(tags[\"Rating\"], 5)\n self.assertIn(\"XMP-mwg-rs:RegionInfo\", tags)\n self.assertIn(\"Alice\", tags[\"XMP-mwg-rs:RegionInfo\"])\n" }, { "path": "api/tests/test_favorite_photos.py", "content": "from unittest.mock import patch\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import create_test_photos, create_test_user\n\n\nclass FavoritePhotosTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user(favorite_min_rating=1)\n self.user2 = create_test_user(favorite_min_rating=1)\n self.client.force_authenticate(user=self.user1)\n\n def test_tag_my_photos_as_favorite(self):\n photos = create_test_photos(number_of_photos=3, owner=self.user1)\n image_hashes = [p.image_hash for p in photos]\n\n payload = {\"image_hashes\": image_hashes, \"favorite\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/favorite/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(3, len(data[\"results\"]))\n self.assertEqual(3, len(data[\"updated\"]))\n self.assertEqual(0, len(data[\"not_updated\"]))\n\n def test_untag_my_photos_as_favorite(self):\n photos1 = create_test_photos(\n number_of_photos=1, owner=self.user1, rating=self.user1.favorite_min_rating\n )\n photos2 = create_test_photos(number_of_photos=2, owner=self.user1)\n image_hashes = [p.image_hash for p in photos1 + photos2]\n\n payload = {\"image_hashes\": image_hashes, \"favorite\": False}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/favorite/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(1, len(data[\"results\"]))\n self.assertEqual(1, len(data[\"updated\"]))\n self.assertEqual(2, len(data[\"not_updated\"]))\n\n def test_tag_photos_of_other_user_as_favorite(self):\n photos = create_test_photos(number_of_photos=2, owner=self.user2)\n image_hashes = [p.image_hash for p in photos]\n\n payload = {\"image_hashes\": image_hashes, \"favorite\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/favorite/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"updated\"]))\n # Photos not owned by user are treated as \"missing\" for security (no info leak)\n self.assertEqual(0, len(data[\"not_updated\"]))\n\n @patch(\"api.views.photos.logger.warning\", autospec=True)\n def test_tag_nonexistent_photo_as_favorite(self, logger):\n payload = {\"image_hashes\": [\"nonexistent_photo\"], \"favorite\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/favorite/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"updated\"]))\n self.assertEqual(0, len(data[\"not_updated\"]))\n logger.assert_called_with(\n \"Could not set photo nonexistent_photo to favorite. It does not exist or is not owned by user.\"\n )\n" }, { "path": "api/tests/test_file_model.py", "content": "import importlib.machinery\nimport importlib.util\nimport sys\nimport types\nimport unittest\nfrom unittest.mock import patch\n\n\ndef _ensure_stub_modules():\n if \"django\" not in sys.modules:\n django_module = types.ModuleType(\"django\")\n django_db_module = types.ModuleType(\"django.db\")\n django_db_models_module = types.ModuleType(\"django.db.models\")\n\n class DummyModel:\n pass\n\n def dummy_field(*args, **kwargs):\n return None\n\n django_db_models_module.Model = DummyModel\n django_db_models_module.CharField = dummy_field\n django_db_models_module.TextField = dummy_field\n django_db_models_module.PositiveIntegerField = dummy_field\n django_db_models_module.BooleanField = dummy_field\n django_db_models_module.ManyToManyField = dummy_field\n\n django_module.db = django_db_module\n django_db_module.models = django_db_models_module\n django_module.__path__ = []\n django_db_module.__path__ = []\n django_module.__spec__ = importlib.machinery.ModuleSpec(\n \"django\", loader=None, is_package=True\n )\n django_db_module.__spec__ = importlib.machinery.ModuleSpec(\n \"django.db\", loader=None, is_package=True\n )\n django_db_models_module.__spec__ = importlib.machinery.ModuleSpec(\n \"django.db.models\", loader=None\n )\n\n sys.modules[\"django\"] = django_module\n sys.modules[\"django.db\"] = django_db_module\n sys.modules[\"django.db.models\"] = django_db_models_module\n\n if \"magic\" not in sys.modules:\n magic_module = types.ModuleType(\"magic\")\n\n class Magic:\n def __init__(self, *args, **kwargs):\n pass\n\n def from_file(self, path):\n return \"application/octet-stream\"\n\n magic_module.Magic = Magic\n magic_module.__spec__ = importlib.machinery.ModuleSpec(\"magic\", loader=None)\n sys.modules[\"magic\"] = magic_module\n\n if \"pyvips\" not in sys.modules:\n pyvips_module = types.ModuleType(\"pyvips\")\n\n class Image:\n @staticmethod\n def thumbnail(*args, **kwargs):\n raise NotImplementedError\n\n class Enums:\n class Size:\n DOWN = \"down\"\n\n pyvips_module.Image = Image\n pyvips_module.enums = Enums\n pyvips_module.__spec__ = importlib.machinery.ModuleSpec(\n \"pyvips\", loader=None\n )\n sys.modules[\"pyvips\"] = pyvips_module\n\n if \"api\" not in sys.modules:\n api_module = types.ModuleType(\"api\")\n api_module.__path__ = []\n api_module.__spec__ = importlib.machinery.ModuleSpec(\n \"api\", loader=None, is_package=True\n )\n\n util_module = types.ModuleType(\"api.util\")\n\n class Logger:\n def error(self, *args, **kwargs):\n pass\n\n util_module.logger = Logger()\n util_module.__spec__ = importlib.machinery.ModuleSpec(\"api.util\", loader=None)\n util_module.__file__ = \"\"\n\n models_module = types.ModuleType(\"api.models\")\n models_module.__path__ = []\n models_module.__spec__ = importlib.machinery.ModuleSpec(\n \"api.models\", loader=None, is_package=True\n )\n\n api_module.util = util_module\n api_module.models = models_module\n\n sys.modules[\"api\"] = api_module\n sys.modules[\"api.util\"] = util_module\n sys.modules[\"api.models\"] = models_module\n\n if \"exiftool\" not in sys.modules:\n exiftool_module = types.ModuleType(\"exiftool\")\n exiftool_module.__spec__ = importlib.machinery.ModuleSpec(\n \"exiftool\", loader=None\n )\n sys.modules[\"exiftool\"] = exiftool_module\n\n if \"requests\" not in sys.modules:\n requests_module = types.ModuleType(\"requests\")\n requests_module.__spec__ = importlib.machinery.ModuleSpec(\n \"requests\", loader=None\n )\n sys.modules[\"requests\"] = requests_module\n\n if \"django.conf\" not in sys.modules:\n django_conf_module = types.ModuleType(\"django.conf\")\n django_conf_module.settings = types.SimpleNamespace(LOGS_ROOT=\"/tmp\")\n django_conf_module.__spec__ = importlib.machinery.ModuleSpec(\n \"django.conf\", loader=None, is_package=True\n )\n sys.modules[\"django.conf\"] = django_conf_module\n\n\ndef _load_file_module():\n _ensure_stub_modules()\n # Use a unique module name to avoid conflicting with Django's registry\n module_name = \"api.models.file._test_stub\"\n if module_name in sys.modules:\n return sys.modules[module_name]\n spec = importlib.util.spec_from_file_location(\n module_name, \"api/models/file.py\"\n )\n module = importlib.util.module_from_spec(spec)\n sys.modules[module_name] = module\n spec.loader.exec_module(module)\n return module\n\n\nclass TestIsVideo(unittest.TestCase):\n @classmethod\n def setUpClass(cls):\n # Only load the module when this specific test class runs\n # Skip if Django models are already loaded to avoid double registration\n if \"api.models.file\" in sys.modules:\n cls._file_module = sys.modules[\"api.models.file\"]\n else:\n cls._file_module = _load_file_module()\n\n def test_is_video_returns_false_when_magic_raises(self):\n class FailingMagic:\n def from_file(self, path):\n raise RuntimeError(\"magic failure\")\n\n with patch.object(\n self._file_module.magic, \"Magic\", return_value=FailingMagic()\n ):\n self.assertFalse(self._file_module.is_video(\"/tmp/test.mp4\"))\n" }, { "path": "api/tests/test_file_path_uniqueness.py", "content": "\"\"\"\nTests for File path uniqueness enforcement.\n\nTests verify:\n- Database unique constraint on File.path\n- File.create() get_or_create pattern\n- Migration deduplication logic\n- Concurrent scan handling\n\"\"\"\n\nimport os\nimport tempfile\nimport threading\n\nfrom django.db import IntegrityError, transaction\nfrom django.test import TestCase, TransactionTestCase\n\nfrom api.models.file import File\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass FilePathUniqueConstraintTestCase(TestCase):\n \"\"\"Tests for the unique constraint on File.path.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_unique_constraint_prevents_duplicate_paths(self):\n \"\"\"Test that the database prevents creating two Files with the same path.\"\"\"\n path = \"/photos/test_image.jpg\"\n \n # Create first file\n _file1 = File.objects.create(\n hash=\"hash1\" + \"a\" * 28,\n path=path,\n type=File.IMAGE,\n )\n \n # Attempting to create second file with same path should fail\n with self.assertRaises(IntegrityError):\n with transaction.atomic():\n File.objects.create(\n hash=\"hash2\" + \"b\" * 28,\n path=path,\n type=File.IMAGE,\n )\n\n def test_unique_constraint_allows_different_paths(self):\n \"\"\"Test that different paths are allowed.\"\"\"\n file1 = File.objects.create(\n hash=\"hash1\" + \"a\" * 28,\n path=\"/photos/image1.jpg\",\n type=File.IMAGE,\n )\n file2 = File.objects.create(\n hash=\"hash2\" + \"b\" * 28,\n path=\"/photos/image2.jpg\",\n type=File.IMAGE,\n )\n \n self.assertEqual(File.objects.count(), 2)\n self.assertNotEqual(file1.path, file2.path)\n\n def test_empty_paths_are_unique(self):\n \"\"\"Test that empty paths are subject to unique constraint.\"\"\"\n # First empty path file\n _file1 = File.objects.create(\n hash=\"hash1\" + \"a\" * 28,\n path=\"\",\n type=File.IMAGE,\n )\n \n # Second empty path should fail\n with self.assertRaises(IntegrityError):\n with transaction.atomic():\n File.objects.create(\n hash=\"hash2\" + \"b\" * 28,\n path=\"\",\n type=File.IMAGE,\n )\n\n\nclass FileCreateMethodTestCase(TestCase):\n \"\"\"Tests for File.create() get_or_create pattern.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n # Create a temp directory for test files\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n # Clean up temp files\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def _create_test_file(self, filename, content=b\"test content\"):\n \"\"\"Helper to create a test file on disk.\"\"\"\n path = os.path.join(self.temp_dir, filename)\n with open(path, \"wb\") as f:\n f.write(content)\n return path\n\n def test_create_returns_existing_file_for_same_path(self):\n \"\"\"Test that File.create() returns existing File for same path.\"\"\"\n path = self._create_test_file(\"test.jpg\")\n \n # Create first file\n file1 = File.create(path, self.user)\n \n # Create second file with same path - should return existing\n file2 = File.create(path, self.user)\n \n # Should be the same file\n self.assertEqual(file1.hash, file2.hash)\n self.assertEqual(file1.path, file2.path)\n \n # Should only have one File in database\n self.assertEqual(File.objects.filter(path=path).count(), 1)\n\n def test_create_creates_new_file_for_different_path(self):\n \"\"\"Test that File.create() creates new File for different path.\"\"\"\n path1 = self._create_test_file(\"test1.jpg\", b\"content1\")\n path2 = self._create_test_file(\"test2.jpg\", b\"content2\")\n \n file1 = File.create(path1, self.user)\n file2 = File.create(path2, self.user)\n \n self.assertNotEqual(file1.hash, file2.hash)\n self.assertNotEqual(file1.path, file2.path)\n self.assertEqual(File.objects.count(), 2)\n\n def test_create_returns_existing_even_if_content_changed(self):\n \"\"\"Test that File.create() returns existing File even if content changed.\"\"\"\n path = self._create_test_file(\"test.jpg\", b\"original content\")\n \n # Create first file\n file1 = File.create(path, self.user)\n original_hash = file1.hash\n \n # Modify file content\n with open(path, \"wb\") as f:\n f.write(b\"modified content\")\n \n # Create again - should return existing File (not recalculate hash)\n file2 = File.create(path, self.user)\n \n # Should return existing file (hash stays the same)\n self.assertEqual(file1.hash, file2.hash)\n self.assertEqual(file2.hash, original_hash)\n\n def test_create_determines_correct_file_type(self):\n \"\"\"Test that File.create() correctly determines file type.\"\"\"\n # Create image file\n img_path = self._create_test_file(\"photo.jpg\")\n img_file = File.create(img_path, self.user)\n self.assertEqual(img_file.type, File.IMAGE)\n \n # Create RAW file\n raw_path = self._create_test_file(\"photo.CR2\")\n raw_file = File.create(raw_path, self.user)\n self.assertEqual(raw_file.type, File.RAW_FILE)\n \n # Create metadata file\n xmp_path = self._create_test_file(\"photo.xmp\")\n xmp_file = File.create(xmp_path, self.user)\n self.assertEqual(xmp_file.type, File.METADATA_FILE)\n\n\nclass MigrationDeduplicationTestCase(TestCase):\n \"\"\"Tests for the migration deduplication logic.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_deduplication_prefers_non_missing_file(self):\n \"\"\"Test that deduplication logic prefers non-missing files.\n \n This tests the scoring logic that the migration uses.\n \"\"\"\n # Create two files to simulate pre-migration state\n file_missing = File.objects.create(\n hash=\"hash_missing\" + \"a\" * 21,\n path=\"/photos/missing_file.jpg\",\n type=File.IMAGE,\n missing=True,\n )\n file_ok = File.objects.create(\n hash=\"hash_ok\" + \"a\" * 25,\n path=\"/photos/ok_file.jpg\",\n type=File.IMAGE,\n missing=False,\n )\n \n # Scoring logic: non-missing files get +100\n def score_file(f):\n score = 0\n if not f.missing:\n score += 100\n return score\n \n # Non-missing file should have higher score\n self.assertGreater(score_file(file_ok), score_file(file_missing))\n\n def test_deduplication_keeps_file_with_more_photos(self):\n \"\"\"Test that deduplication prefers files linked to more photos.\"\"\"\n _path = \"/photos/popular.jpg\"\n \n # Create two files\n file_popular = File.objects.create(\n hash=\"hash_popular\" + \"a\" * 20,\n path=\"/photos/popular1.jpg\",\n type=File.IMAGE,\n )\n file_lonely = File.objects.create(\n hash=\"hash_lonely\" + \"a\" * 21,\n path=\"/photos/lonely1.jpg\",\n type=File.IMAGE,\n )\n \n # Link popular file to multiple photos\n for i in range(3):\n photo = create_test_photo(owner=self.user)\n photo.files.add(file_popular)\n photo.save()\n \n # Link lonely file to one photo\n photo = create_test_photo(owner=self.user)\n photo.files.add(file_lonely)\n photo.save()\n \n # Verify photo counts\n self.assertEqual(file_popular.photo_set.count(), 3)\n self.assertEqual(file_lonely.photo_set.count(), 1)\n\n\nclass ConcurrentScanTestCase(TransactionTestCase):\n \"\"\"Tests for concurrent scan handling with unique constraint.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def _create_test_file(self, filename, content=b\"test content\"):\n \"\"\"Helper to create a test file on disk.\"\"\"\n path = os.path.join(self.temp_dir, filename)\n with open(path, \"wb\") as f:\n f.write(content)\n return path\n\n def test_concurrent_create_same_path_no_duplicates(self):\n \"\"\"Test that concurrent File.create() calls don't create duplicates.\"\"\"\n path = self._create_test_file(\"concurrent_test.jpg\")\n results = []\n errors = []\n \n def create_file():\n try:\n file = File.create(path, self.user)\n results.append(file.hash)\n except Exception as e:\n errors.append(str(e))\n \n # Run multiple threads trying to create the same file\n threads = []\n for _ in range(5):\n t = threading.Thread(target=create_file)\n threads.append(t)\n \n for t in threads:\n t.start()\n \n for t in threads:\n t.join()\n \n # All should succeed (due to get_or_create pattern)\n self.assertEqual(len(errors), 0, f\"Errors occurred: {errors}\")\n \n # All should return the same file\n self.assertTrue(len(set(results)) <= 1, \n f\"Expected same hash for all, got: {results}\")\n \n # Should only have one File in database\n self.assertEqual(File.objects.filter(path=path).count(), 1)\n\n def test_concurrent_create_different_paths_succeeds(self):\n \"\"\"Test that concurrent creates of different paths all succeed.\"\"\"\n paths = [\n self._create_test_file(f\"concurrent_test_{i}.jpg\", f\"content{i}\".encode())\n for i in range(5)\n ]\n results = []\n \n def create_file(path):\n file = File.create(path, self.user)\n results.append(file.hash)\n \n threads = []\n for path in paths:\n t = threading.Thread(target=create_file, args=(path,))\n threads.append(t)\n \n for t in threads:\n t.start()\n \n for t in threads:\n t.join()\n \n # All 5 files should be created\n self.assertEqual(len(results), 5)\n self.assertEqual(File.objects.count(), 5)\n\n\nclass FilePathLookupTestCase(TestCase):\n \"\"\"Tests for path-based lookups.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_filter_by_path_is_exact(self):\n \"\"\"Test that filtering by path is exact match.\"\"\"\n file1 = File.objects.create(\n hash=\"hash1\" + \"a\" * 28,\n path=\"/photos/image.jpg\",\n type=File.IMAGE,\n )\n _file2 = File.objects.create(\n hash=\"hash2\" + \"b\" * 28,\n path=\"/photos/image2.jpg\",\n type=File.IMAGE,\n )\n \n # Exact match should find only one\n result = File.objects.filter(path=\"/photos/image.jpg\")\n self.assertEqual(result.count(), 1)\n self.assertEqual(result.first().hash, file1.hash)\n\n def test_photo_files_path_lookup(self):\n \"\"\"Test that Photo.files.filter(path=...) works correctly.\"\"\"\n file1 = File.objects.create(\n hash=\"hash1\" + \"a\" * 28,\n path=\"/photos/image1.jpg\",\n type=File.IMAGE,\n )\n file2 = File.objects.create(\n hash=\"hash2\" + \"b\" * 28,\n path=\"/photos/image2.jpg\",\n type=File.IMAGE,\n )\n \n photo = create_test_photo(owner=self.user)\n photo.files.add(file1, file2)\n \n # Should find exact path\n self.assertTrue(photo.files.filter(path=\"/photos/image1.jpg\").exists())\n self.assertFalse(photo.files.filter(path=\"/photos/image3.jpg\").exists())\n\n\nclass PhotoFileAssociationTestCase(TestCase):\n \"\"\"Tests for Photo-File associations with unique path constraint.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_multiple_photos_can_share_same_file(self):\n \"\"\"Test that multiple Photos can reference the same File.\"\"\"\n file = File.objects.create(\n hash=\"shared_hash\" + \"a\" * 23,\n path=\"/photos/shared_image.jpg\",\n type=File.IMAGE,\n )\n \n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n photo1.files.add(file)\n photo1.main_file = file\n photo1.save()\n \n photo2.files.add(file)\n photo2.main_file = file\n photo2.save()\n \n # Both photos should reference the same file\n self.assertEqual(photo1.main_file.hash, photo2.main_file.hash)\n self.assertEqual(file.photo_set.count(), 2)\n\n def test_photo_with_multiple_file_variants(self):\n \"\"\"Test Photo with multiple file variants (JPEG + RAW).\"\"\"\n jpeg_file = File.objects.create(\n hash=\"jpeg_hash\" + \"a\" * 24,\n path=\"/photos/image.jpg\",\n type=File.IMAGE,\n )\n raw_file = File.objects.create(\n hash=\"raw_hash\" + \"a\" * 25,\n path=\"/photos/image.CR2\",\n type=File.RAW_FILE,\n )\n\n photo = create_test_photo(owner=self.user)\n photo.files.add(jpeg_file, raw_file)\n photo.main_file = jpeg_file\n photo.save()\n\n # Photo should have both explicitly added files\n # Note: create_test_photo sets main_file but doesn't add it to photo.files\n self.assertEqual(photo.files.count(), 2)\n self.assertTrue(photo.files.filter(path=\"/photos/image.jpg\").exists())\n self.assertTrue(photo.files.filter(path=\"/photos/image.CR2\").exists())\n" }, { "path": "api/tests/test_geocode.py", "content": "from unittest.mock import patch\n\nfrom constance.test import override_config\nfrom django.test import TestCase\n\nfrom api.geocode.geocode import reverse_geocode\nfrom api.geocode.parsers.mapbox import parse as parse_mapbox\nfrom api.geocode.parsers.nominatim import parse as parse_nominatim\nfrom api.geocode.parsers.opencage import parse as parse_opencage\nfrom api.geocode.parsers.tomtom import parse as parse_tomtom\nfrom api.tests.fixtures.geocode.expectations.mapbox import (\n expectations as mapbox_expectations,\n)\nfrom api.tests.fixtures.geocode.expectations.nominatim import (\n expectations as nominatim_expectations,\n)\nfrom api.tests.fixtures.geocode.expectations.opencage import (\n expectations as opencage_expectations,\n)\nfrom api.tests.fixtures.geocode.expectations.tomtom import (\n expectations as tomtom_expectations,\n)\nfrom api.tests.fixtures.geocode.responses.mapbox import responses as mapbox_responses\nfrom api.tests.fixtures.geocode.responses.nominatim import (\n responses as nominatim_responses,\n)\nfrom api.tests.fixtures.geocode.responses.opencage import (\n responses as opencage_responses,\n)\nfrom api.tests.fixtures.geocode.responses.tomtom import responses as tomtom_responses\n\n\nclass MapboxLocation:\n def __init__(self, raw):\n self.raw = raw\n self.address = raw[\"place_name\"]\n\n\nclass TomTomLocation:\n def __init__(self, raw):\n self.raw = raw\n self.address = raw[\"address\"][\"freeformAddress\"]\n\n\nclass NominatimLocation:\n def __init__(self, raw):\n self.raw = raw\n self.address = raw[\"display_name\"]\n\n\nclass OpenCageLocation:\n def __init__(self, raw):\n self.raw = raw\n self.address = raw[\"formatted\"]\n\n\nclass TestGeocodeParsers(TestCase):\n def test_mapbox_parser(self):\n for index, raw in enumerate(mapbox_responses):\n self.assertEqual(\n parse_mapbox(MapboxLocation(raw)), mapbox_expectations[index]\n )\n\n def test_tomtom_parser(self):\n for index, raw in enumerate(tomtom_responses):\n self.assertEqual(\n parse_tomtom(TomTomLocation(raw)), tomtom_expectations[index]\n )\n\n def test_nominatim_parser(self):\n for index, raw in enumerate(nominatim_responses):\n self.assertEqual(\n parse_nominatim(NominatimLocation(raw)), nominatim_expectations[index]\n )\n\n def test_opencage_parser(self):\n for index, raw in enumerate(opencage_responses):\n self.assertEqual(\n parse_opencage(OpenCageLocation(raw)), opencage_expectations[index]\n )\n\n\nclass FakeLocation:\n raw = None\n address = None\n\n def __init__(self, location):\n self.raw = location\n self.address = location[\"place_name\"]\n\n\nclass FakeProvider:\n def __init__(self, response):\n self.response = response\n\n def reverse(self, _):\n return FakeLocation(self.response)\n\n\ndef fake_geocoder(response):\n return lambda **_: FakeProvider(response)\n\n\nclass TestGeocoder(TestCase):\n @override_config(MAP_API_PROVIDER=\"mapbox\")\n @patch(\"geopy.get_geocoder_for_service\", autospec=True)\n def test_reverse_geocode(self, get_geocoder_for_service_mock):\n get_geocoder_for_service_mock.return_value = fake_geocoder(mapbox_responses[1])\n result = reverse_geocode(0, 0)\n self.assertEqual(result, mapbox_expectations[1])\n\n @override_config(MAP_API_PROVIDER=\"mapbox\")\n @override_config(MAP_API_KEY=\"\")\n def test_reverse_geocode_no_api_key(self):\n result = reverse_geocode(0, 0)\n self.assertEqual(result, {})\n" }, { "path": "api/tests/test_get_faces.py", "content": "from django.test import TestCase\nfrom django.urls import reverse\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import (\n create_test_face,\n create_test_person,\n create_test_photo,\n create_test_user,\n)\n\n\nclass IncompleteFacesTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n self.photo = create_test_photo(owner=self.user)\n\n def test_if_classification_person_is_ignored_if_below_threshold(self):\n \"\"\"Test if unknown faces with classification are returned. Only classification_probability and min_confidence should be looked into.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo,\n classification_person=person,\n classification_probability=0.4,\n )\n\n response = self.client.get(\n reverse(\"incomplete_faces-list\"),\n {\n \"inferred\": \"true\",\n \"analysis_method\": \"classification\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data), 1)\n self.assertIn(\n \"Unknown - Other\", response.data[0][\"name\"]\n ) # Ensure unknown face is returned\n\n def test_if_min_confidence_and_prob_are_compared_correctly(\n self,\n ):\n \"\"\"Test that incomplete faces with classification analysis method are returned properly.\"\"\"\n create_test_face(\n photo=self.photo,\n classification_probability=0.3,\n )\n create_test_face(\n photo=self.photo,\n classification_probability=0.4,\n )\n create_test_face(\n photo=self.photo,\n classification_probability=0.6,\n )\n\n response = self.client.get(\n reverse(\"incomplete_faces-list\"),\n {\n \"inferred\": \"true\",\n \"analysis_method\": \"classification\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertIn(\n \"Unknown - Other\", response.data[0][\"name\"]\n ) # Ensure unknown face is returned\n # face count should be number of faces with classification probability less than 0.5\n self.assertEqual(response.data[0][\"face_count\"], 2)\n\n def test_incomplete_faces_with_clustering(self):\n \"\"\"Test that incomplete faces with clustering analysis method are returned properly.\"\"\"\n create_test_face(\n photo=self.photo,\n classification_person=None,\n classification_probability=0.5,\n cluster_person=None,\n cluster_probability=0.8,\n )\n create_test_face(\n photo=self.photo,\n classification_person=None,\n classification_probability=0.5,\n cluster_person=None,\n cluster_probability=0.4,\n )\n\n response = self.client.get(\n reverse(\"incomplete_faces-list\"),\n {\n \"inferred\": \"true\",\n \"analysis_method\": \"clustering\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertIn(\"Unknown - Other\", response.data[0][\"name\"])\n # face count should be number of faces with clustering person = None\n self.assertEqual(response.data[0][\"face_count\"], 2)\n\n def test_no_inferred_faces(self):\n \"\"\"Test when there are no inferred faces and only user-labeled faces should appear.\"\"\"\n person = create_test_person(name=\"John Doe\", cluster_owner=self.user)\n create_test_face(photo=self.photo, person=person)\n\n response = self.client.get(\n reverse(\"incomplete_faces-list\"), {\"inferred\": \"false\"}\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertIn(\n \"John Doe\", response.data[0][\"name\"]\n ) # Ensure user-labeled face is returned\n\n\nclass FaceListViewTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n self.photo = create_test_photo(owner=self.user)\n\n def test_min_confidence_when_classification(self):\n \"\"\"Test that faces with classification are returned properly.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo,\n classification_person=person,\n classification_probability=0.6,\n )\n create_test_face(\n photo=self.photo,\n classification_person=person,\n classification_probability=0.4,\n )\n\n response = self.client.get(\n reverse(\"faces-list\"),\n {\n \"person\": person.id,\n \"analysis_method\": \"classification\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data[\"results\"]), 1)\n\n def test_min_confidence_but_for_unknown_other(self):\n \"\"\"Test that unknown faces with classification are returned properly.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo,\n classification_person=person,\n classification_probability=0.4,\n )\n\n response = self.client.get(\n reverse(\"faces-list\"),\n {\n \"person\": \"0\",\n \"analysis_method\": \"classification\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data[\"results\"]), 1)\n self.assertEqual(0.4, response.data[\"results\"][0][\"person_label_probability\"])\n\n def test_min_confidence_when_clustering(self):\n \"\"\"Test that faces with clustering are returned properly.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo, cluster_person=person, cluster_probability=0.6\n )\n create_test_face(\n photo=self.photo, cluster_person=person, cluster_probability=0.4\n )\n\n response = self.client.get(\n reverse(\"faces-list\"),\n {\n \"person\": person.id,\n \"analysis_method\": \"clustering\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data[\"results\"]), 1)\n\n def test_min_confidence_when_clustering_and_unknown(self):\n \"\"\"Test that unknown faces with clustering are returned properly.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo, cluster_person=person, cluster_probability=0.4\n )\n\n response = self.client.get(\n reverse(\"faces-list\"),\n {\n \"person\": \"0\",\n \"analysis_method\": \"clustering\",\n \"min_confidence\": \"0.5\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertEqual(len(response.data[\"results\"]), 1)\n self.assertEqual(0.4, response.data[\"results\"][0][\"person_label_probability\"])\n\n def test_face_list_classification_order_by_probability(self):\n \"\"\"Test that faces with classification are ordered by classification probability.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo,\n classification_person=person,\n classification_probability=0.7,\n )\n create_test_face(\n photo=self.photo,\n classification_person=person,\n classification_probability=0.9,\n )\n\n response = self.client.get(\n reverse(\"faces-list\"),\n {\n \"person\": person.id,\n \"analysis_method\": \"classification\",\n \"order_by\": \"probability\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertGreater(\n response.data[\"results\"][0][\"person_label_probability\"],\n response.data[\"results\"][1][\"person_label_probability\"],\n )\n\n def test_face_list_clustering_order_by_probability(self):\n \"\"\"Test that faces with clustering are ordered by clustering probability.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n create_test_face(\n photo=self.photo, cluster_person=person, cluster_probability=0.6\n )\n create_test_face(\n photo=self.photo, cluster_person=person, cluster_probability=0.9\n )\n\n response = self.client.get(\n reverse(\"faces-list\"),\n {\n \"person\": person.id,\n \"analysis_method\": \"clustering\",\n \"order_by\": \"probability\",\n },\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertGreater(\n response.data[\"results\"][0][\"person_label_probability\"],\n response.data[\"results\"][1][\"person_label_probability\"],\n )\n\n def test_face_list_order_by_date(self):\n \"\"\"Test that faces can be ordered by the photo's timestamp when 'order_by' is set to 'date'.\"\"\"\n person = create_test_person(cluster_owner=self.user)\n photo = create_test_photo(\n owner=self.user, exif_timestamp=\"2021-01-01T00:00:00Z\"\n )\n photo2 = create_test_photo(\n owner=self.user, exif_timestamp=\"2021-01-02T00:00:00Z\"\n )\n create_test_face(photo=photo, person=person)\n create_test_face(photo=photo2, person=person)\n response = self.client.get(\n reverse(\"faces-list\"),\n {\"person\": person.id, \"inferred\": False, \"order_by\": \"date\"},\n )\n\n self.assertEqual(response.status_code, 200)\n self.assertLess(\n response.data[\"results\"][0][\"timestamp\"],\n response.data[\"results\"][1][\"timestamp\"],\n )\n" }, { "path": "api/tests/test_hide_photos.py", "content": "from unittest.mock import patch\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import create_test_photos, create_test_user\n\n\nclass FavoritePhotosTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n def test_hide_my_photos(self):\n photos = create_test_photos(number_of_photos=3, owner=self.user1)\n image_hashes = [p.image_hash for p in photos]\n\n payload = {\"image_hashes\": image_hashes, \"hidden\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/hide/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(3, len(data[\"results\"]))\n self.assertEqual(3, len(data[\"updated\"]))\n self.assertEqual(0, len(data[\"not_updated\"]))\n\n def test_untag_my_photos_as_favorite(self):\n photos1 = create_test_photos(number_of_photos=1, owner=self.user1, hidden=True)\n photos2 = create_test_photos(number_of_photos=2, owner=self.user1)\n image_hashes = [p.image_hash for p in photos1 + photos2]\n\n payload = {\"image_hashes\": image_hashes, \"hidden\": False}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/hide/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(1, len(data[\"results\"]))\n self.assertEqual(1, len(data[\"updated\"]))\n self.assertEqual(2, len(data[\"not_updated\"]))\n\n def test_tag_photos_of_other_user_as_favorite(self):\n photos = create_test_photos(number_of_photos=2, owner=self.user2)\n image_hashes = [p.image_hash for p in photos]\n\n payload = {\"image_hashes\": image_hashes, \"hidden\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/hide/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"updated\"]))\n # Photos not owned by user are treated as \"missing\" for security (no info leak)\n self.assertEqual(0, len(data[\"not_updated\"]))\n\n @patch(\"api.views.photos.logger.warning\", autospec=True)\n def test_tag_nonexistent_photo_as_favorite(self, logger):\n payload = {\"image_hashes\": [\"nonexistent_photo\"], \"hidden\": True}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/hide/\", format=\"json\", data=payload, headers=headers\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n self.assertEqual(0, len(data[\"results\"]))\n self.assertEqual(0, len(data[\"updated\"]))\n self.assertEqual(0, len(data[\"not_updated\"]))\n logger.assert_called_with(\n \"Could not set photo nonexistent_photo to hidden. It does not exist or is not owned by user.\"\n )\n" }, { "path": "api/tests/test_im2txt.py", "content": "import json\nimport os\nimport statistics\nimport threading\nimport time\nfrom unittest import skip\n\nimport psutil\nimport torch\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.image_captioning import generate_caption, unload_model\nfrom api.tests.utils import create_test_user\nfrom api.util import logger\n\n\ndef test_coco(testcase, device=\"cpu\", model=\"im2txt\"):\n from pycocoevalcap.eval import COCOEvalCap\n from pycocotools.coco import COCO\n\n logger.info(f\"{device} {model}\")\n blip = model == \"blip\"\n\n # Path to the annotations file\n annotation_file = (\n os.path.dirname(os.path.abspath(__file__))\n + \"/fixtures/coco/captions_val2017.json\"\n )\n\n # Initialize COCO API for annotations\n coco_caps = COCO(annotation_file)\n\n # Get a list of all image IDs in the dataset\n all_image_ids = coco_caps.getImgIds()\n\n # Load all images using the COCO API\n all_images = coco_caps.loadImgs(all_image_ids)\n\n # Save the generated captions together with the image IDs\n generated_captions = []\n counter = 0\n # Iterate through the valdation images\n for image_info in all_images:\n image = (\n os.path.dirname(os.path.abspath(__file__))\n + \"/fixtures/coco/val2017/\"\n + image_info[\"file_name\"]\n )\n caption = generate_caption(blip=blip, image_path=image)\n # The LSTM adds a and token to the generated caption\n caption = caption.replace(\"\", \"\").replace(\"\", \"\").strip().lower()\n generated_captions.append({\"image_id\": image_info[\"id\"], \"caption\": caption})\n counter += 1\n if counter > 100000:\n break\n\n testcase.end_time = time.time()\n\n # Define the path to the output JSON file\n output_json_file = (\n os.path.dirname(os.path.abspath(__file__))\n + \"/fixtures/coco/\"\n + \"generated_captions.json\"\n )\n\n # Write the generated_captions to the JSON file\n with open(output_json_file, \"w\") as json_file:\n json.dump(generated_captions, json_file)\n\n # create coco object and coco_result object\n coco = COCO(annotation_file)\n coco_result = coco.loadRes(output_json_file)\n\n # create coco_eval object by taking coco and coco_result\n coco_eval = COCOEvalCap(coco, coco_result)\n\n # evaluate on a subset of images by setting\n coco_eval.params[\"image_id\"] = coco_result.getImgIds()\n\n # evaluate results\n # SPICE will take a few minutes the first time, but speeds up due to caching\n coco_eval.evaluate()\n\n # print output evaluation scores\n for metric, score in coco_eval.eval.items():\n logger.info(f\"{metric}: {score:.3f}\")\n\n\nclass Im2TxtBenchmark(TestCase):\n gpu_available = torch.cuda.is_available()\n # Start monitoring RAM usage\n process = psutil.Process(os.getpid())\n start_ram_usage = process.memory_info().rss\n # Start measuring timepycocoevalcap\n start_time = time.time()\n ram_usages = [] # List to store RAM usage samples\n ram_monitor_thread = None\n end_time = None\n\n def setUp(self) -> None:\n # Check if the required files exist in the fixtures directory\n self.coco_fixture_path = os.path.join(self.fixtures_dir, \"coco\")\n self.val2017_fixture_path = os.path.join(self.fixtures_dir, \"coco\", \"val2017\")\n self.val2017captions_fixture_path = os.path.join(\n self.fixtures_dir, \"coco\", \"captions_val2017.json\"\n )\n\n if not os.path.exists(self.coco_fixture_path):\n logger.warning(\n f\"Skipping tests. Directory not found: {self.coco_fixture_path}. Please add a coco folder to the fixtures directory.\"\n )\n self.skipTest(\"Directory not found\")\n\n if not os.path.exists(self.val2017_fixture_path):\n logger.warning(\n f\"Skipping tests. Directory not found: {self.val2017_fixture_path}. Validation images are required for the COCO benchmark. Please download the COCO validation images and place them in the fixtures/coco/val2017 directory.\"\n )\n self.skipTest(\"Directory not found\")\n if not os.path.exists(self.val2017captions_fixture_path):\n logger.warning(\n f\"Skipping tests. Directory not found: {self.val2017captions_fixture_path}. Captions of Validation images are required for the COCO benchmark. Please download the COCO validation images and place them in the fixtures/coco/ directory.\"\n )\n self.skipTest(\"Directory not found\")\n unload_model()\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n self.gpu_available = torch.cuda.is_available()\n self.process = psutil.Process(os.getpid())\n self.start_ram_usage = self.process.memory_info().rss\n self.start_time = time.time()\n self.ram_usages = [] # List to store RAM usage samples\n self.ram_monitor_thread = threading.Thread(\n target=self.monitor_ram_usage, args=(self.process, self.ram_usages, 5)\n ) # Monitor RAM for 5 seconds\n self.ram_monitor_thread.start()\n\n def tearDown(self) -> None:\n if self.end_time is None:\n execution_time = time.time() - self.start_time\n else:\n execution_time = self.end_time - self.start_time\n self.end_time = None\n self.ram_monitor_thread.join()\n # Calculate RAM usage statistics\n mean_ram_usage_mb = statistics.mean(self.ram_usages) / (\n 1024 * 1024\n ) # Convert bytes to MB\n median_ram_usage_mb = statistics.median(self.ram_usages) / (\n 1024 * 1024\n ) # Convert bytes to MB\n\n # Log the results using the logger function\n logger.info(\"Test Result:\")\n logger.info(f\"GPU Used: {'Yes' if self.gpu_available else 'No'}\")\n logger.info(f\"Mean RAM Usage: {mean_ram_usage_mb:.2f} MB\")\n logger.info(f\"Median RAM Usage: {median_ram_usage_mb:.2f} MB\")\n logger.info(f\"Execution Time: {execution_time:.2f} seconds\")\n\n def monitor_ram_usage(self, process, ram_usages, duration):\n # Monitor RAM usage for the specified duration\n start_time = time.time()\n while time.time() - start_time < duration:\n ram_usage = process.memory_info().rss\n ram_usages.append(ram_usage)\n time.sleep(0.1) # Poll every 1 second\n\n @skip\n def test_im2txt_cpu(self):\n file = os.path.dirname(os.path.abspath(__file__)) + \"/fixtures/niaz.jpg\"\n self.gpu_available = \"False\"\n caption = generate_caption(device=torch.device(\"cpu\"), image_path=file)\n\n self.assertEqual(\n \" a man with a beard is holding a remote control . \", caption\n )\n\n logger.info(f\"Caption: {caption}\")\n\n @skip\n def test_im2txt_gpu(self):\n file = os.path.dirname(os.path.abspath(__file__)) + \"/fixtures/niaz.jpg\"\n caption = generate_caption(device=torch.device(\"cuda\"), image_path=file)\n\n self.assertEqual(\n \" a man with a beard is holding a remote control . \", caption\n )\n\n logger.info(f\"Caption: {caption}\")\n\n @skip\n def test_im2txt_cpu_100(self):\n file = os.path.dirname(os.path.abspath(__file__)) + \"/fixtures/niaz.jpg\"\n self.gpu_available = \"False\"\n\n for i in range(100):\n caption = generate_caption(device=torch.device(\"cpu\"), image_path=file)\n self.assertEqual(\n \" a man with a beard is holding a remote control . \",\n caption,\n )\n\n @skip\n def test_im2txt_gpu_100(self):\n file = os.path.dirname(os.path.abspath(__file__)) + \"/fixtures/niaz.jpg\"\n\n for i in range(100):\n caption = generate_caption(device=torch.device(\"cuda\"), image_path=file)\n self.assertEqual(\n \" a man with a beard is holding a remote control . \",\n caption,\n )\n\n @skip\n def test_im2txt_coco_cpu(self):\n test_coco(testcase=self, device=\"cpu\", model=\"im2txt\")\n\n @skip\n def test_im2txt_coco_gpu(self):\n test_coco(testcase=self, device=\"cuda\", model=\"im2txt\")\n\n @skip\n def test_blip_coco_cpu(self):\n test_coco(self, \"cpu\", \"blip\")\n\n @skip\n def test_blip_coco_gpu(self):\n test_coco(self, \"cuda\", \"blip\")\n" }, { "path": "api/tests/test_live_photo.py", "content": "\"\"\"\nComprehensive tests for api/stacks/live_photo.py\n\nTests the Live Photo detection and stacking logic:\n- Google Pixel Motion Photo detection (embedded MP4 after JPEG EOI)\n- Samsung Motion Photo detection (MotionPhoto_Data marker)\n- Apple Live Photo detection (paired .mov file)\n- Stack creation for detected live photos\n- Batch processing\n\"\"\"\n\nimport os\nimport tempfile\nfrom pathlib import Path\nfrom unittest.mock import MagicMock, Mock, patch\n\nfrom django.test import TestCase, override_settings\n\nfrom api.models.file import File\nfrom api.models.photo import Photo\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.user import User\nfrom api.stacks.live_photo import (\n APPLE_LIVE_PHOTO_EXTENSIONS,\n GOOGLE_PIXEL_MP4_SIGNATURES,\n JPEG_EOI_MARKER,\n SAMSUNG_MOTION_MARKER,\n _create_apple_live_photo_stack,\n _create_embedded_live_photo_stack,\n _locate_google_embedded_video,\n _locate_samsung_embedded_video,\n detect_live_photo,\n extract_embedded_motion_video,\n find_apple_live_photo_video,\n has_embedded_motion_video,\n process_live_photos_batch,\n)\n\n\nclass LocateGoogleEmbeddedVideoTestCase(TestCase):\n \"\"\"Tests for the _locate_google_embedded_video function.\"\"\"\n\n def test_finds_ftypmp42_signature(self):\n \"\"\"Should find MP4 with ftypmp42 signature.\"\"\"\n # Build data: JPEG content + 4 padding bytes + ftyp signature\n data = b\"JPEG_CONTENT\\xff\\xd9\" + b\"\\x00\\x00\\x00\\x00\" + b\"ftypmp42\" + b\"more_video_data\"\n position = _locate_google_embedded_video(data)\n expected = data.find(b\"ftypmp42\") - 4\n self.assertEqual(position, expected)\n\n def test_finds_ftypisom_signature(self):\n \"\"\"Should find MP4 with ftypisom signature.\"\"\"\n data = b\"JPEG_CONTENT\\xff\\xd9\" + b\"\\x00\\x00\\x00\\x20\" + b\"ftypisom\"\n position = _locate_google_embedded_video(data)\n expected = data.find(b\"ftypisom\") - 4\n self.assertEqual(position, expected)\n\n def test_finds_ftypiso2_signature(self):\n \"\"\"Should find MP4 with ftypiso2 signature.\"\"\"\n data = b\"JPEG_CONTENT\\xff\\xd9\" + b\"\\x00\\x00\\x00\\x20\" + b\"ftypiso2\"\n position = _locate_google_embedded_video(data)\n expected = data.find(b\"ftypiso2\") - 4\n self.assertEqual(position, expected)\n\n def test_returns_minus_one_when_not_found(self):\n \"\"\"Should return -1 if no signature found.\"\"\"\n data = b\"JPEG_CONTENT\\xff\\xd9_no_video_here\"\n position = _locate_google_embedded_video(data)\n self.assertEqual(position, -1)\n\n def test_empty_data(self):\n \"\"\"Should return -1 for empty data.\"\"\"\n position = _locate_google_embedded_video(b\"\")\n self.assertEqual(position, -1)\n\n def test_finds_first_signature_if_multiple(self):\n \"\"\"Should find the first signature if multiple exist.\"\"\"\n data = b\"JPEG\" + b\"\\x00\\x00\\x00\\x00\" + b\"ftypmp42\" + b\"MIDDLE\" + b\"ftypisom\"\n position = _locate_google_embedded_video(data)\n # Should find ftypmp42 first\n self.assertEqual(position, 4) # 4 bytes for \"JPEG\"\n\n\nclass LocateSamsungEmbeddedVideoTestCase(TestCase):\n \"\"\"Tests for the _locate_samsung_embedded_video function.\"\"\"\n\n def test_finds_samsung_marker(self):\n \"\"\"Should find Samsung motion photo marker.\"\"\"\n data = b\"JPEG_CONTENT\\xff\\xd9\" + SAMSUNG_MOTION_MARKER + b\"video_data\"\n position = _locate_samsung_embedded_video(data)\n expected = data.find(SAMSUNG_MOTION_MARKER) + len(SAMSUNG_MOTION_MARKER)\n self.assertEqual(position, expected)\n\n def test_returns_minus_one_when_not_found(self):\n \"\"\"Should return -1 if no marker found.\"\"\"\n data = b\"JPEG_CONTENT\\xff\\xd9_no_motion_marker\"\n position = _locate_samsung_embedded_video(data)\n self.assertEqual(position, -1)\n\n def test_empty_data(self):\n \"\"\"Should return -1 for empty data.\"\"\"\n position = _locate_samsung_embedded_video(b\"\")\n self.assertEqual(position, -1)\n\n\nclass HasEmbeddedMotionVideoTestCase(TestCase):\n \"\"\"Tests for the has_embedded_motion_video function.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test files.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n @patch(\"api.stacks.live_photo.magic.Magic\")\n def test_returns_false_for_non_jpeg(self, mock_magic_class):\n \"\"\"Should return False for non-JPEG files.\"\"\"\n mock_magic = MagicMock()\n mock_magic.from_file.return_value = \"image/png\"\n mock_magic_class.return_value = mock_magic\n\n result = has_embedded_motion_video(\"/some/path.png\")\n self.assertFalse(result)\n\n @patch(\"api.stacks.live_photo.magic.Magic\")\n @patch(\"builtins.open\")\n @patch(\"api.stacks.live_photo.mmap\")\n def test_returns_true_for_google_motion_photo(self, mock_mmap, mock_open, mock_magic_class):\n \"\"\"Should return True for Google Motion Photo.\"\"\"\n mock_magic = MagicMock()\n mock_magic.from_file.return_value = \"image/jpeg\"\n mock_magic_class.return_value = mock_magic\n\n # Mock file data with Google MP4 signature\n mock_data = b\"JPEG\" + b\"\\x00\\x00\\x00\\x00\" + b\"ftypmp42\"\n mock_mm = MagicMock()\n mock_mm.__enter__ = Mock(return_value=mock_data)\n mock_mm.__exit__ = Mock(return_value=False)\n mock_mmap.return_value = mock_mm\n\n mock_file = MagicMock()\n mock_file.__enter__ = Mock(return_value=mock_file)\n mock_file.__exit__ = Mock(return_value=False)\n mock_open.return_value = mock_file\n\n with patch(\"api.stacks.live_photo._locate_google_embedded_video\", return_value=4):\n result = has_embedded_motion_video(\"/some/path.jpg\")\n self.assertTrue(result)\n\n @patch(\"api.stacks.live_photo.magic.Magic\")\n def test_returns_false_on_exception(self, mock_magic_class):\n \"\"\"Should return False and log warning on exception.\"\"\"\n mock_magic_class.side_effect = Exception(\"File not found\")\n\n with patch(\"api.stacks.live_photo.logger\") as mock_logger:\n result = has_embedded_motion_video(\"/nonexistent/path.jpg\")\n self.assertFalse(result)\n mock_logger.warning.assert_called()\n\n\nclass FindAppleLivePhotoVideoTestCase(TestCase):\n \"\"\"Tests for the find_apple_live_photo_video function.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test files.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_finds_lowercase_mov_companion(self):\n \"\"\"Should find .mov companion file.\"\"\"\n # Create test files\n image_path = os.path.join(self.temp_dir, \"IMG_001.jpg\")\n video_path = os.path.join(self.temp_dir, \"IMG_001.mov\")\n Path(image_path).touch()\n Path(video_path).touch()\n\n result = find_apple_live_photo_video(image_path)\n self.assertEqual(result, video_path)\n\n def test_finds_uppercase_mov_companion(self):\n \"\"\"Should find .MOV companion file (uppercase).\"\"\"\n image_path = os.path.join(self.temp_dir, \"IMG_002.HEIC\")\n video_path = os.path.join(self.temp_dir, \"IMG_002.MOV\")\n Path(image_path).touch()\n Path(video_path).touch()\n\n result = find_apple_live_photo_video(image_path)\n self.assertEqual(result, video_path)\n\n def test_returns_none_when_no_companion(self):\n \"\"\"Should return None if no companion video exists.\"\"\"\n image_path = os.path.join(self.temp_dir, \"IMG_003.jpg\")\n Path(image_path).touch()\n\n result = find_apple_live_photo_video(image_path)\n self.assertIsNone(result)\n\n def test_prefers_lowercase_mov(self):\n \"\"\"Should prefer .mov over .MOV if both exist.\"\"\"\n image_path = os.path.join(self.temp_dir, \"IMG_004.jpg\")\n video_lowercase = os.path.join(self.temp_dir, \"IMG_004.mov\")\n video_uppercase = os.path.join(self.temp_dir, \"IMG_004.MOV\")\n Path(image_path).touch()\n Path(video_lowercase).touch()\n Path(video_uppercase).touch()\n\n result = find_apple_live_photo_video(image_path)\n # Should find .mov first (lowercase is first in APPLE_LIVE_PHOTO_EXTENSIONS)\n self.assertEqual(result, video_lowercase)\n\n def test_handles_different_image_extensions(self):\n \"\"\"Should work with various image extensions.\"\"\"\n for ext in [\".jpg\", \".JPG\", \".heic\", \".HEIC\", \".jpeg\"]:\n image_path = os.path.join(self.temp_dir, f\"test{ext}\")\n video_path = os.path.join(self.temp_dir, \"test.mov\")\n Path(image_path).touch()\n Path(video_path).touch()\n\n result = find_apple_live_photo_video(image_path)\n self.assertEqual(result, video_path)\n\n # Cleanup for next iteration\n Path(image_path).unlink()\n Path(video_path).unlink()\n\n\nclass ExtractEmbeddedMotionVideoTestCase(TestCase):\n \"\"\"Tests for the extract_embedded_motion_video function.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test files.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n @override_settings(MEDIA_ROOT=None)\n def test_extracts_google_motion_video(self):\n \"\"\"Should extract embedded MP4 from Google Motion Photo.\"\"\"\n # Use temp_dir as MEDIA_ROOT\n with self.settings(MEDIA_ROOT=self.temp_dir):\n # Create a fake motion photo file\n fake_video_data = b\"fake_mp4_video_content\"\n file_content = (\n b\"JPEG_IMAGE_DATA\\xff\\xd9\" + # JPEG with EOI marker\n b\"\\x00\\x00\\x00\\x00\" + # 4 padding bytes\n b\"ftypmp42\" + # ftyp signature\n fake_video_data\n )\n\n input_path = os.path.join(self.temp_dir, \"motion_photo.jpg\")\n with open(input_path, \"wb\") as f:\n f.write(file_content)\n\n result = extract_embedded_motion_video(input_path, \"test_hash_123\")\n\n self.assertIsNotNone(result)\n self.assertIn(\"test_hash_123_motion.mp4\", result)\n self.assertTrue(os.path.exists(result))\n\n # Verify extracted content starts from ftyp\n with open(result, \"rb\") as f:\n extracted = f.read()\n self.assertTrue(extracted.startswith(b\"\\x00\\x00\\x00\\x00ftypmp42\"))\n\n def test_returns_none_for_no_embedded_video(self):\n \"\"\"Should return None if no embedded video found.\"\"\"\n with self.settings(MEDIA_ROOT=self.temp_dir):\n # Create a regular JPEG without embedded video\n input_path = os.path.join(self.temp_dir, \"regular.jpg\")\n with open(input_path, \"wb\") as f:\n f.write(b\"JPEG_DATA\\xff\\xd9\")\n\n result = extract_embedded_motion_video(input_path, \"hash123\")\n self.assertIsNone(result)\n\n def test_returns_none_on_file_error(self):\n \"\"\"Should return None and log error if file access fails.\"\"\"\n with patch(\"api.stacks.live_photo.logger\") as mock_logger:\n result = extract_embedded_motion_video(\"/nonexistent/file.jpg\", \"hash\")\n self.assertIsNone(result)\n mock_logger.error.assert_called()\n\n\nclass DetectLivePhotoTestCase(TestCase):\n \"\"\"Tests for the detect_live_photo function.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and temporary files.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n self.user = User.objects.create(username=\"livetest\")\n\n def tearDown(self):\n \"\"\"Clean up.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_returns_none_for_photo_without_main_file(self):\n \"\"\"Should return None if photo has no main_file.\"\"\"\n photo = MagicMock()\n photo.main_file = None\n\n result = detect_live_photo(photo, self.user)\n self.assertIsNone(result)\n\n @patch(\"api.stacks.live_photo.has_embedded_motion_video\")\n @patch(\"api.stacks.live_photo._create_embedded_live_photo_stack\")\n def test_detects_embedded_motion_video(self, mock_create, mock_has_embedded):\n \"\"\"Should detect and create stack for embedded motion video.\"\"\"\n mock_has_embedded.return_value = True\n mock_stack = MagicMock()\n mock_create.return_value = mock_stack\n\n photo = MagicMock()\n photo.main_file.path = \"/path/to/image.jpg\"\n\n result = detect_live_photo(photo, self.user)\n\n mock_create.assert_called_once_with(photo, self.user)\n self.assertEqual(result, mock_stack)\n\n @patch(\"api.stacks.live_photo.has_embedded_motion_video\")\n @patch(\"api.stacks.live_photo.find_apple_live_photo_video\")\n @patch(\"api.stacks.live_photo._create_apple_live_photo_stack\")\n def test_detects_apple_live_photo(self, mock_create, mock_find, mock_has_embedded):\n \"\"\"Should detect and create stack for Apple Live Photo.\"\"\"\n mock_has_embedded.return_value = False\n mock_find.return_value = \"/path/to/video.mov\"\n mock_stack = MagicMock()\n mock_create.return_value = mock_stack\n\n photo = MagicMock()\n photo.main_file.path = \"/path/to/image.jpg\"\n\n result = detect_live_photo(photo, self.user)\n\n mock_create.assert_called_once_with(photo, \"/path/to/video.mov\", self.user)\n self.assertEqual(result, mock_stack)\n\n @patch(\"api.stacks.live_photo.has_embedded_motion_video\")\n @patch(\"api.stacks.live_photo.find_apple_live_photo_video\")\n def test_returns_none_for_regular_photo(self, mock_find, mock_has_embedded):\n \"\"\"Should return None for regular photos without motion.\"\"\"\n mock_has_embedded.return_value = False\n mock_find.return_value = None\n\n photo = MagicMock()\n photo.main_file.path = \"/path/to/regular.jpg\"\n\n result = detect_live_photo(photo, self.user)\n self.assertIsNone(result)\n\n\nclass CreateEmbeddedLivePhotoStackTestCase(TestCase):\n \"\"\"Tests for _create_embedded_live_photo_stack function.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user.\"\"\"\n self.user = User.objects.create(username=\"embedtest\")\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n @override_settings(FEATURE_PROCESS_EMBEDDED_MEDIA=False)\n def test_returns_none_if_feature_disabled(self):\n \"\"\"Should return None if embedded media processing is disabled.\"\"\"\n photo = MagicMock()\n\n with patch(\"api.stacks.live_photo.logger\") as mock_logger:\n result = _create_embedded_live_photo_stack(photo, self.user)\n self.assertIsNone(result)\n mock_logger.debug.assert_called()\n\n @override_settings(FEATURE_PROCESS_EMBEDDED_MEDIA=True)\n @patch(\"api.stacks.live_photo.extract_embedded_motion_video\")\n def test_returns_none_if_extraction_fails(self, mock_extract):\n \"\"\"Should return None if video extraction fails.\"\"\"\n mock_extract.return_value = None\n\n photo = MagicMock()\n photo.main_file.path = \"/path/to/image.jpg\"\n photo.main_file.hash = \"abc123\"\n\n result = _create_embedded_live_photo_stack(photo, self.user)\n self.assertIsNone(result)\n\n @override_settings(FEATURE_PROCESS_EMBEDDED_MEDIA=True)\n @patch(\"api.stacks.live_photo.extract_embedded_motion_video\")\n @patch(\"api.stacks.live_photo.File.create\")\n def test_returns_existing_stack_if_present(self, mock_file_create, mock_extract):\n \"\"\"Should return existing stack if photo already has one.\"\"\"\n mock_extract.return_value = \"/path/to/video.mp4\"\n mock_video_file = MagicMock()\n mock_file_create.return_value = mock_video_file\n\n # Create a mock photo with existing stack\n existing_stack = MagicMock()\n photo = MagicMock()\n photo.main_file.path = \"/path/to/image.jpg\"\n photo.main_file.hash = \"abc123\"\n photo.main_file.embedded_media = MagicMock()\n photo.stacks.filter.return_value.first.return_value = existing_stack\n\n result = _create_embedded_live_photo_stack(photo, self.user)\n self.assertEqual(result, existing_stack)\n\n\nclass CreateAppleLivePhotoStackTestCase(TestCase):\n \"\"\"Tests for _create_apple_live_photo_stack function.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user and file.\"\"\"\n from django.utils import timezone\n\n self.user = User.objects.create(username=\"appletest\")\n self.temp_dir = tempfile.mkdtemp()\n\n # Create a test file\n self.file_hash = \"a\" * 32\n self.file_path = os.path.join(self.temp_dir, \"test.jpg\")\n Path(self.file_path).touch()\n\n self.file = File.objects.create(\n hash=self.file_hash,\n path=self.file_path,\n type=File.IMAGE,\n )\n\n # Create a test photo with required fields\n self.photo = Photo.objects.create(\n owner=self.user,\n main_file=self.file,\n image_hash=\"b\" * 32,\n added_on=timezone.now(),\n )\n\n def tearDown(self):\n \"\"\"Clean up.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_creates_new_stack_for_apple_live_photo(self):\n \"\"\"Should create new Live Photo stack.\"\"\"\n video_path = os.path.join(self.temp_dir, \"test.mov\")\n Path(video_path).touch()\n\n # Create a real video file to avoid foreign key issues\n video_file = File.objects.create(\n hash=\"c\" * 32,\n path=video_path,\n type=File.VIDEO,\n )\n\n with patch(\"api.stacks.live_photo.File.create\", return_value=video_file):\n with patch(\"api.stacks.live_photo.File.objects.filter\") as mock_filter:\n # Simulate video file not existing yet\n mock_filter.return_value.first.return_value = None\n\n result = _create_apple_live_photo_stack(self.photo, video_path, self.user)\n\n self.assertIsNotNone(result)\n self.assertEqual(result.stack_type, PhotoStack.StackType.LIVE_PHOTO)\n self.assertEqual(result.primary_photo, self.photo)\n self.assertEqual(result.owner, self.user)\n\n def test_returns_existing_stack_if_present(self):\n \"\"\"Should return existing stack if photo already has one.\"\"\"\n # Create existing stack\n existing_stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.LIVE_PHOTO,\n primary_photo=self.photo,\n )\n self.photo.stacks.add(existing_stack)\n\n video_path = os.path.join(self.temp_dir, \"test.mov\")\n Path(video_path).touch()\n\n # Create a real video file\n video_file = File.objects.create(\n hash=\"d\" * 32,\n path=video_path,\n type=File.VIDEO,\n )\n\n with patch(\"api.stacks.live_photo.File.create\", return_value=video_file):\n with patch(\"api.stacks.live_photo.File.objects.filter\") as mock_filter:\n mock_filter.return_value.first.return_value = None\n\n result = _create_apple_live_photo_stack(self.photo, video_path, self.user)\n\n self.assertEqual(result, existing_stack)\n\n\nclass ProcessLivePhotosBatchTestCase(TestCase):\n \"\"\"Tests for the process_live_photos_batch function.\"\"\"\n\n def setUp(self):\n \"\"\"Create test user.\"\"\"\n self.user = User.objects.create(username=\"batchtest\")\n\n @patch(\"api.stacks.live_photo.detect_live_photo\")\n def test_processes_all_photos(self, mock_detect):\n \"\"\"Should process all photos in the list.\"\"\"\n mock_detect.return_value = None\n\n photos = [MagicMock() for _ in range(5)]\n result = process_live_photos_batch(self.user, photos)\n\n self.assertEqual(mock_detect.call_count, 5)\n self.assertEqual(result[\"detected\"], 0)\n self.assertEqual(result[\"stacks_created\"], 0)\n\n @patch(\"api.stacks.live_photo.detect_live_photo\")\n def test_counts_detected_live_photos(self, mock_detect):\n \"\"\"Should count detected live photos.\"\"\"\n mock_stack = MagicMock()\n mock_stack.photo_count = 2 # Existing stack with photos\n mock_detect.side_effect = [mock_stack, None, mock_stack, None]\n\n photos = [MagicMock() for _ in range(4)]\n result = process_live_photos_batch(self.user, photos)\n\n self.assertEqual(result[\"detected\"], 2)\n\n @patch(\"api.stacks.live_photo.detect_live_photo\")\n def test_counts_new_stacks_created(self, mock_detect):\n \"\"\"Should count newly created stacks.\"\"\"\n new_stack = MagicMock()\n new_stack.photo_count = 1 # New stack (just the photo)\n mock_detect.return_value = new_stack\n\n photos = [MagicMock() for _ in range(3)]\n result = process_live_photos_batch(self.user, photos)\n\n self.assertEqual(result[\"detected\"], 3)\n self.assertEqual(result[\"stacks_created\"], 3)\n\n @patch(\"api.stacks.live_photo.detect_live_photo\")\n def test_handles_exceptions_gracefully(self, mock_detect):\n \"\"\"Should continue processing after exceptions.\"\"\"\n mock_detect.side_effect = [Exception(\"Error\"), MagicMock(photo_count=2)]\n\n photos = [MagicMock(), MagicMock()]\n photos[0].id = \"photo1\"\n photos[1].id = \"photo2\"\n\n with patch(\"api.stacks.live_photo.logger\") as mock_logger:\n result = process_live_photos_batch(self.user, photos)\n mock_logger.error.assert_called()\n self.assertEqual(result[\"detected\"], 1)\n\n def test_empty_list_returns_zero_counts(self):\n \"\"\"Should return zero counts for empty list.\"\"\"\n result = process_live_photos_batch(self.user, [])\n self.assertEqual(result[\"detected\"], 0)\n self.assertEqual(result[\"stacks_created\"], 0)\n\n\nclass ConstantsTestCase(TestCase):\n \"\"\"Tests for module constants.\"\"\"\n\n def test_jpeg_eoi_marker(self):\n \"\"\"JPEG EOI marker should be correct.\"\"\"\n self.assertEqual(JPEG_EOI_MARKER, b\"\\xff\\xd9\")\n\n def test_google_signatures_list(self):\n \"\"\"Google MP4 signatures should be defined.\"\"\"\n self.assertIn(b\"ftypmp42\", GOOGLE_PIXEL_MP4_SIGNATURES)\n self.assertIn(b\"ftypisom\", GOOGLE_PIXEL_MP4_SIGNATURES)\n self.assertIn(b\"ftypiso2\", GOOGLE_PIXEL_MP4_SIGNATURES)\n\n def test_samsung_marker(self):\n \"\"\"Samsung motion marker should be correct.\"\"\"\n self.assertEqual(SAMSUNG_MOTION_MARKER, b\"MotionPhoto_Data\")\n\n def test_apple_extensions(self):\n \"\"\"Apple Live Photo extensions should include .mov variants.\"\"\"\n self.assertIn(\".mov\", APPLE_LIVE_PHOTO_EXTENSIONS)\n self.assertIn(\".MOV\", APPLE_LIVE_PHOTO_EXTENSIONS)\n\n\nclass EdgeCasesTestCase(TestCase):\n \"\"\"Edge case tests for live photo detection.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up.\"\"\"\n import shutil\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_unicode_filename_apple_live_photo(self):\n \"\"\"Should handle unicode characters in filenames.\"\"\"\n image_path = os.path.join(self.temp_dir, \"照片_日本_🌸.jpg\")\n video_path = os.path.join(self.temp_dir, \"照片_日本_🌸.mov\")\n Path(image_path).touch()\n Path(video_path).touch()\n\n result = find_apple_live_photo_video(image_path)\n self.assertEqual(result, video_path)\n\n def test_special_characters_in_path(self):\n \"\"\"Should handle special characters in file path.\"\"\"\n image_path = os.path.join(self.temp_dir, \"test photo (1).jpg\")\n video_path = os.path.join(self.temp_dir, \"test photo (1).mov\")\n Path(image_path).touch()\n Path(video_path).touch()\n\n result = find_apple_live_photo_video(image_path)\n self.assertEqual(result, video_path)\n\n def test_locate_video_at_start_of_data(self):\n \"\"\"Should handle video marker at very start of data.\"\"\"\n data = b\"ftypmp42rest_of_video\"\n position = _locate_google_embedded_video(data)\n self.assertEqual(position, -4) # Would be negative, which is fine\n\n def test_multiple_samsung_markers(self):\n \"\"\"Should find first Samsung marker if multiple present.\"\"\"\n data = (\n SAMSUNG_MOTION_MARKER + b\"first_video\" +\n SAMSUNG_MOTION_MARKER + b\"second_video\"\n )\n position = _locate_samsung_embedded_video(data)\n self.assertEqual(position, len(SAMSUNG_MOTION_MARKER))\n\n def test_partial_signature_not_matched(self):\n \"\"\"Should not match partial signatures.\"\"\"\n # ftypm instead of ftypmp42\n data = b\"JPEG\\xff\\xd9\\x00\\x00\\x00\\x00ftypm\"\n position = _locate_google_embedded_video(data)\n self.assertEqual(position, -1)\n\n def test_very_large_file_simulation(self):\n \"\"\"Should handle large data efficiently.\"\"\"\n # Create 10MB of fake data with marker near end\n large_data = b\"A\" * (10 * 1024 * 1024)\n large_data += b\"\\x00\\x00\\x00\\x00ftypmp42\"\n\n position = _locate_google_embedded_video(large_data)\n self.assertGreater(position, 0)\n\n def test_binary_data_with_nulls(self):\n \"\"\"Should handle binary data with null bytes.\"\"\"\n data = b\"\\x00\" * 100 + b\"\\x00\\x00\\x00\\x00ftypmp42\" + b\"\\x00\" * 50\n position = _locate_google_embedded_video(data)\n self.assertEqual(position, 100)\n" }, { "path": "api/tests/test_location_timeline.py", "content": "import csv\nimport os\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.stats import get_location_timeline, get_photo_month_counts\nfrom api.models import Photo\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\ndef prepare_database(user):\n data = (\n os.path.dirname(os.path.abspath(__file__))\n + \"/fixtures/location_timeline_test_data.csv\"\n )\n with open(data) as f:\n reader = csv.reader(f)\n for row in reader:\n if row[0].startswith(\"#\"):\n continue\n country = row[0]\n exif_timestamp = row[1]\n geolocation_json = {\"places\": [country], \"features\": [{\"text\": country}]}\n create_test_photo(\n owner=user,\n exif_timestamp=exif_timestamp,\n geolocation_json=geolocation_json,\n )\n\n\nexpected_location_timeline = [\n {\n \"data\": [22208418.0],\n \"color\": \"#a6cee3\",\n \"loc\": \"Germany\",\n \"start\": 1576286343.0,\n \"end\": 1598494761.0,\n },\n {\n \"data\": [9413609.0],\n \"color\": \"#1f78b4\",\n \"loc\": \"Canada\",\n \"start\": 1598494761.0,\n \"end\": 1607908370.0,\n },\n {\n \"data\": [20648022.0],\n \"color\": \"#b2df8a\",\n \"loc\": \"France\",\n \"start\": 1607908370.0,\n \"end\": 1628556392.0,\n },\n {\n \"data\": [6132785.0],\n \"color\": \"#33a02c\",\n \"loc\": \"Canada\",\n \"start\": 1628556392.0,\n \"end\": 1634689177.0,\n },\n {\n \"data\": [79828.0],\n \"color\": \"#fb9a99\",\n \"loc\": \"France\",\n \"start\": 1634689177.0,\n \"end\": 1634769005.0,\n },\n]\n\nexpected_photo_month_counts = [\n {\"month\": \"2019-12\", \"count\": 4},\n {\"month\": \"2020-1\", \"count\": 0},\n {\"month\": \"2020-2\", \"count\": 0},\n {\"month\": \"2020-3\", \"count\": 0},\n {\"month\": \"2020-4\", \"count\": 0},\n {\"month\": \"2020-5\", \"count\": 0},\n {\"month\": \"2020-6\", \"count\": 0},\n {\"month\": \"2020-7\", \"count\": 0},\n {\"month\": \"2020-8\", \"count\": 4},\n {\"month\": \"2020-9\", \"count\": 0},\n {\"month\": \"2020-10\", \"count\": 0},\n {\"month\": \"2020-11\", \"count\": 0},\n {\"month\": \"2020-12\", \"count\": 4},\n {\"month\": \"2021-1\", \"count\": 0},\n {\"month\": \"2021-2\", \"count\": 0},\n {\"month\": \"2021-3\", \"count\": 0},\n {\"month\": \"2021-4\", \"count\": 0},\n {\"month\": \"2021-5\", \"count\": 0},\n {\"month\": \"2021-6\", \"count\": 0},\n {\"month\": \"2021-7\", \"count\": 0},\n {\"month\": \"2021-8\", \"count\": 4},\n {\"month\": \"2021-9\", \"count\": 0},\n {\"month\": \"2021-10\", \"count\": 4},\n]\n\n\nclass LocationTimelineTest(TestCase):\n def setUp(self) -> None:\n self.user = create_test_user()\n self.client = APIClient()\n self.client.force_authenticate(user=self.user)\n Photo.objects.all().delete()\n prepare_database(self.user)\n\n def test_location_timeline_endpoint(self):\n response = self.client.get(\"/api/locationtimeline/\")\n result = response.json()\n self.assertEqual(result, expected_location_timeline)\n\n def test_get_location_timeline(self):\n result = get_location_timeline(self.user)\n self.assertEqual(result, expected_location_timeline)\n\n def test_get_photo_month_counts_endpoint(self):\n response = self.client.get(\"/api/photomonthcounts/\")\n result = response.json()\n self.assertEqual(result, expected_photo_month_counts)\n\n def test_get_photo_month_count(self):\n result = get_photo_month_counts(self.user)\n self.assertEqual(result, expected_photo_month_counts)\n" }, { "path": "api/tests/test_metadata_ordering_sentinel.py", "content": "import os\nimport random\nimport tempfile\nimport uuid\nfrom unittest.mock import patch\n\nfrom django.test import TestCase, override_settings\n\nfrom api.models import Photo\nfrom api.tests.utils import create_test_user\n\n\ndef create_unique_png(seed=0):\n \"\"\"\n Generate a minimal valid PNG with unique content based on seed.\n Each different seed produces a different hash.\n \"\"\"\n import struct\n\n def png_chunk(chunk_type, data):\n chunk_data = chunk_type + data\n crc = 0xFFFFFFFF\n for byte in chunk_data:\n crc ^= byte\n for _ in range(8):\n crc = (crc >> 1) ^ 0xEDB88320 if crc & 1 else crc >> 1\n crc ^= 0xFFFFFFFF\n return struct.pack(\">I\", len(data)) + chunk_data + struct.pack(\">I\", crc)\n\n # PNG signature\n png_sig = b\"\\x89PNG\\r\\n\\x1a\\n\"\n\n # IHDR: 1x1 image, 8-bit RGB\n ihdr = struct.pack(\">IIBBBBB\", 1, 1, 8, 2, 0, 0, 0)\n\n # IDAT: compressed image data with seed-based variation\n idat_data = bytes([seed % 256]) + b\"\\x00\\x00\\x00\\x00\\x00\"\n import zlib\n\n idat_compressed = zlib.compress(idat_data)\n\n # IEND: end of PNG\n iend = b\"\"\n\n return (\n png_sig\n + png_chunk(b\"IHDR\", ihdr)\n + png_chunk(b\"IDAT\", idat_compressed)\n + png_chunk(b\"IEND\", iend)\n )\n\n\nclass DummyAsyncTask:\n \"\"\"Synchronous replacement for django_q.tasks.AsyncTask.\n\n - Immediately executes the callable.\n - Tracks completion counts per group id when used for image tasks.\n \"\"\"\n\n GROUP_COMPLETIONS: dict[str, int] = {}\n\n def __init__(self, func, *args, **kwargs):\n self.func = func\n self.args = args\n # Extract 'group' from kwargs before passing to func (func doesn't accept it)\n self.group_id = kwargs.pop(\"group\", None)\n self.kwargs = kwargs\n\n def run(self):\n # Execute the callable synchronously (without 'group' in kwargs)\n result = self.func(*self.args, **self.kwargs)\n\n # If this was an image/video task scheduled with a group,\n # increment the completion counter for that group\n func_name = getattr(self.func, \"__name__\", \"\")\n if self.group_id and func_name == \"handle_new_image\":\n DummyAsyncTask.GROUP_COMPLETIONS[self.group_id] = (\n DummyAsyncTask.GROUP_COMPLETIONS.get(self.group_id, 0) + 1\n )\n return result\n\n\nclass DummyChain:\n def __init__(self, *args, **kwargs):\n self.appended = []\n\n def append(self, *args, **kwargs):\n self.appended.append((args, kwargs))\n return self\n\n def run(self):\n return None\n\n\nclass MetadataOrderingSentinelTest(TestCase):\n def test_random_order_images_and_xmp_are_consistently_linked(self):\n user = create_test_user()\n with tempfile.TemporaryDirectory() as tmpdir:\n user.scan_directory = tmpdir\n user.save(update_fields=[\"scan_directory\"])\n\n # Create N image files and corresponding XMP sidecars\n N = 4\n image_paths = []\n xmp_paths = []\n for i in range(N):\n base = f\"img_{i}\"\n img_path = os.path.join(tmpdir, f\"{base}.jpg\")\n xmp_path = os.path.join(tmpdir, f\"{base}.xmp\")\n with open(img_path, \"wb\") as f:\n f.write(create_unique_png(i)) # Each image has unique hash\n with open(xmp_path, \"wb\") as f:\n f.write(b\"test\")\n image_paths.append(img_path)\n xmp_paths.append(xmp_path)\n\n # Randomize processing order explicitly via scan_files\n all_files = image_paths + xmp_paths\n random.shuffle(all_files)\n\n # Patch environment to make processing synchronous and lightweight\n with override_settings(MEDIA_ROOT=tmpdir):\n with (\n patch(\"api.directory_watcher.scan_jobs.AsyncTask\", DummyAsyncTask),\n patch(\"api.directory_watcher.scan_jobs.Chain\", DummyChain),\n patch(\n \"django_q.tasks.count_group\",\n side_effect=lambda gid: DummyAsyncTask.GROUP_COMPLETIONS.get(\n gid, 0\n ),\n ),\n patch(\n \"api.directory_watcher.scan_jobs.db.connections.close_all\"\n ) as _close_all,\n patch(\n \"api.directory_watcher.scan_jobs.update_scan_counter\"\n ) as _update_counter,\n patch(\"api.directory_watcher.scan_jobs.util.logger\") as _logger,\n patch(\"pyvips.Image.thumbnail\") as _thumb,\n patch(\n \"api.models.thumbnail.Thumbnail._generate_thumbnail\"\n ) as _gen_thumb,\n patch(\n \"api.models.thumbnail.Thumbnail._calculate_aspect_ratio\"\n ) as _calc_ar,\n patch(\n \"api.models.thumbnail.Thumbnail._get_dominant_color\"\n ) as _dom_color,\n patch(\n \"api.models.photo_metadata.PhotoMetadata.extract_exif_data\"\n ) as _exif,\n patch(\n \"api.models.photo.Photo._extract_date_time_from_exif\"\n ) as _exif_dt,\n ):\n # No-op patches\n _thumb.return_value = None\n _close_all.return_value = None\n _update_counter.side_effect = lambda *_args, **_kwargs: None\n _logger.info.side_effect = lambda *_a, **_k: None\n _logger.warning.side_effect = lambda *_a, **_k: None\n _logger.exception.side_effect = lambda *_a, **_k: None\n _gen_thumb.return_value = None\n _calc_ar.return_value = None\n _dom_color.return_value = None\n _exif.return_value = None\n _exif_dt.return_value = None\n\n job_id = str(uuid.uuid4())\n # Emulate the core of scan_photos sequencing explicitly:\n # 1) Enqueue all images/videos in a group and run them synchronously\n # 2) Run the sentinel to process metadata after the group completes\n from api.directory_watcher import (\n handle_new_image,\n wait_for_group_and_process_metadata,\n )\n\n image_group_id = str(uuid.uuid4())\n for img in image_paths:\n DummyAsyncTask(\n handle_new_image, user, img, job_id, group=image_group_id\n ).run()\n\n DummyAsyncTask(\n wait_for_group_and_process_metadata,\n image_group_id,\n xmp_paths,\n user.id,\n False,\n job_id,\n len(image_paths),\n ).run()\n\n # Validate: image tasks ran and each image must have its XMP associated to the same Photo\n total_completions = sum(DummyAsyncTask.GROUP_COMPLETIONS.values())\n self.assertEqual(\n total_completions,\n N,\n msg=f\"Expected {N} image task completions, got {total_completions}\",\n )\n\n photos = list(Photo.objects.all())\n self.assertEqual(\n len(photos), N, msg=\"All images should produce Photo objects\"\n )\n\n # Build a map from image base name to whether an XMP is linked\n linked = {}\n for p in photos:\n # main_file.path is the image path\n main_path = p.main_file.path if p.main_file else \"\"\n base = os.path.splitext(os.path.basename(main_path))[0]\n xmp_list = list(\n p.files.filter(path__endswith=\".xmp\").values_list(\"path\", flat=True)\n )\n linked[base] = len(xmp_list) >= 1\n\n # All should be True\n self.assertTrue(\n all(linked.values()), msg=f\"Some photos missing XMP: {linked}\"\n )\n" }, { "path": "api/tests/test_migration_0099.py", "content": "\"\"\"\nTests for migration 0099_photo_uuid_primary_key.\n\nVerifies the UUID primary key migration works correctly on both\nPostgreSQL (raw SQL) and SQLite (table recreation pattern).\n\nStrategy:\n- Migration 0099 is irreversible, so we cannot use the standard\n \"roll back → seed → migrate forward\" pattern.\n- Instead we build a standalone in-memory SQLite database with the\n pre-migration schema, seed data, run the migration function\n directly, and verify the result (TestSQLiteMigration0099).\n- We also verify the post-migration schema on the live Django test DB\n (which already had 0099 applied during test-database creation)\n in TestPostMigrationSchema.\n- Helper functions are tested in isolation in TestSQLiteHelpers.\n\"\"\"\n\nimport sqlite3\nimport uuid\nfrom importlib import import_module\nfrom unittest.mock import MagicMock, patch\n\nfrom django.db import connection\nfrom django.test import TestCase, TransactionTestCase\n\n# Import the migration module (name starts with a digit, use importlib)\n_mod = import_module(\"api.migrations.0099_photo_uuid_primary_key\")\n\n\n# ============================================================================\n# Helpers\n# ============================================================================\n\ndef _sqlite_table_exists(cursor, table_name):\n cursor.execute(\n \"SELECT COUNT(*) FROM sqlite_master WHERE type='table' AND name=?\",\n [table_name],\n )\n return cursor.fetchone()[0] > 0\n\n\ndef _sqlite_column_names(cursor, table_name):\n cursor.execute(f'PRAGMA table_info(\"{table_name}\")')\n return {row[1] for row in cursor.fetchall()}\n\n\ndef _sqlite_pk_columns(cursor, table_name):\n cursor.execute(f'PRAGMA table_info(\"{table_name}\")')\n return [row[1] for row in cursor.fetchall() if row[5]]\n\n\ndef _sqlite_index_exists(cursor, index_name):\n cursor.execute(\n \"SELECT COUNT(*) FROM sqlite_master WHERE type='index' AND name=?\",\n [index_name],\n )\n return cursor.fetchone()[0] > 0\n\n\n# ============================================================================\n# Pre-migration schema builder (mimics the tables that exist at 0098)\n# ============================================================================\n\n_PRE_MIGRATION_DDL = \"\"\"\nCREATE TABLE api_photo (\n image_hash VARCHAR(64) NOT NULL PRIMARY KEY,\n hidden INTEGER NOT NULL DEFAULT 0,\n rating INTEGER NOT NULL DEFAULT 0,\n deleted INTEGER NOT NULL DEFAULT 0,\n video INTEGER NOT NULL DEFAULT 0\n);\n\nCREATE TABLE api_face (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash),\n location_top INTEGER NOT NULL DEFAULT 0,\n location_bottom INTEGER NOT NULL DEFAULT 0,\n location_left INTEGER NOT NULL DEFAULT 0,\n location_right INTEGER NOT NULL DEFAULT 0,\n deleted INTEGER NOT NULL DEFAULT 0,\n classification_probability REAL NOT NULL DEFAULT 0.0,\n cluster_probability REAL NOT NULL DEFAULT 0.0\n);\n\nCREATE TABLE api_thumbnail (\n photo_id VARCHAR(64) NOT NULL PRIMARY KEY REFERENCES api_photo(image_hash),\n aspect_ratio REAL\n);\n\nCREATE TABLE api_photo_caption (\n photo_id VARCHAR(64) NOT NULL PRIMARY KEY REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_photo_search (\n photo_id VARCHAR(64) NOT NULL PRIMARY KEY REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_person (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n name TEXT NOT NULL,\n kind TEXT NOT NULL DEFAULT 'USER',\n cover_photo_id VARCHAR(64) REFERENCES api_photo(image_hash),\n face_count INTEGER NOT NULL DEFAULT 0\n);\n\nCREATE TABLE api_albumuser (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n title TEXT,\n cover_photo_id VARCHAR(64) REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_photo_shared_to (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash),\n user_id INTEGER NOT NULL\n);\n\nCREATE TABLE api_photo_files (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash),\n file_id INTEGER NOT NULL\n);\n\nCREATE TABLE api_albumuser_photos (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n albumuser_id INTEGER NOT NULL,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_albumthing_photos (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n albumthing_id INTEGER NOT NULL,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_albumplace_photos (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n albumplace_id INTEGER NOT NULL,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_albumdate_photos (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n albumdate_id INTEGER NOT NULL,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_albumauto_photos (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n albumauto_id INTEGER NOT NULL,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_albumthing_cover_photos (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n albumthing_id INTEGER NOT NULL,\n photo_id VARCHAR(64) NOT NULL REFERENCES api_photo(image_hash)\n);\n\nCREATE TABLE api_photostack (\n id TEXT PRIMARY KEY,\n primary_photo_id VARCHAR(64) REFERENCES api_photo(image_hash)\n);\n\nCREATE INDEX api_face_photo_id_old ON api_face(photo_id);\n\"\"\"\n\n\ndef _build_test_db():\n \"\"\"Create an in-memory SQLite DB with the pre-migration schema and seed data.\n\n Returns (connection, hashes) where hashes is a list of image_hash values.\n \"\"\"\n conn = sqlite3.connect(\":memory:\")\n cur = conn.cursor()\n cur.executescript(_PRE_MIGRATION_DDL)\n\n hashes = [f\"hash_{i:032d}\" for i in range(1, 4)]\n for h in hashes:\n cur.execute(\n \"INSERT INTO api_photo (image_hash) VALUES (?)\", [h]\n )\n\n # face referencing photo 0\n cur.execute(\n \"INSERT INTO api_face (photo_id, location_top, location_bottom, \"\n \"location_left, location_right, deleted, classification_probability, \"\n \"cluster_probability) VALUES (?, 0, 100, 0, 100, 0, 0.5, 0.5)\",\n [hashes[0]],\n )\n # second face referencing photo 1\n cur.execute(\n \"INSERT INTO api_face (photo_id, location_top, location_bottom, \"\n \"location_left, location_right, deleted, classification_probability, \"\n \"cluster_probability) VALUES (?, 10, 200, 10, 200, 0, 0.8, 0.3)\",\n [hashes[1]],\n )\n # thumbnail for photo 0\n cur.execute(\"INSERT INTO api_thumbnail (photo_id, aspect_ratio) VALUES (?, 1.5)\", [hashes[0]])\n # caption for photo 1\n cur.execute(\"INSERT INTO api_photo_caption (photo_id) VALUES (?)\", [hashes[1]])\n # search for photo 2\n cur.execute(\"INSERT INTO api_photo_search (photo_id) VALUES (?)\", [hashes[2]])\n # person with cover_photo\n cur.execute(\"INSERT INTO api_person (name, cover_photo_id) VALUES (?, ?)\", [\"Alice\", hashes[0]])\n # album user with cover_photo\n cur.execute(\"INSERT INTO api_albumuser (title, cover_photo_id) VALUES (?, ?)\", [\"My Album\", hashes[1]])\n # M2M entries\n cur.execute(\"INSERT INTO api_photo_shared_to (photo_id, user_id) VALUES (?, 1)\", [hashes[0]])\n cur.execute(\"INSERT INTO api_albumuser_photos (albumuser_id, photo_id) VALUES (1, ?)\", [hashes[0]])\n cur.execute(\"INSERT INTO api_albumthing_photos (albumthing_id, photo_id) VALUES (1, ?)\", [hashes[1]])\n cur.execute(\"INSERT INTO api_albumplace_photos (albumplace_id, photo_id) VALUES (1, ?)\", [hashes[2]])\n cur.execute(\"INSERT INTO api_albumdate_photos (albumdate_id, photo_id) VALUES (1, ?)\", [hashes[0]])\n cur.execute(\"INSERT INTO api_albumauto_photos (albumauto_id, photo_id) VALUES (1, ?)\", [hashes[1]])\n cur.execute(\"INSERT INTO api_albumthing_cover_photos (albumthing_id, photo_id) VALUES (1, ?)\", [hashes[2]])\n cur.execute(\"INSERT INTO api_photostack (id, primary_photo_id) VALUES (?, ?)\", [\"stack-1\", hashes[0]])\n\n conn.commit()\n return conn, hashes\n\n\ndef _run_migration_on(sqlite_conn):\n \"\"\"Run the SQLite migration path on the given raw sqlite3 connection.\"\"\"\n # _migrate_sqlite expects a Django-like schema_editor with\n # .connection.cursor() returning something with .execute/.fetchall.\n # We wrap the raw sqlite3 connection to match.\n class _CursorWrapper:\n \"\"\"Thin adapter so _migrate_sqlite can call cursor.execute(sql, params).\"\"\"\n def __init__(self, raw_cursor):\n self._cur = raw_cursor\n def execute(self, sql, params=None):\n if params is None:\n return self._cur.execute(sql)\n return self._cur.execute(sql, params)\n def fetchall(self):\n return self._cur.fetchall()\n def fetchone(self):\n return self._cur.fetchone()\n\n class _ConnWrapper:\n def __init__(self, raw_conn):\n self._conn = raw_conn\n self.vendor = \"sqlite\"\n def cursor(self):\n return _CursorWrapper(self._conn.cursor())\n\n class _SchemaEditor:\n def __init__(self, raw_conn):\n self.connection = _ConnWrapper(raw_conn)\n\n _mod._migrate_sqlite(_SchemaEditor(sqlite_conn))\n sqlite_conn.commit()\n\n\n# ============================================================================\n# Test: Full end-to-end SQLite migration\n# ============================================================================\n\nclass TestSQLiteMigration0099(TestCase):\n \"\"\"\n End-to-end test of the SQLite migration path.\n\n Creates a standalone in-memory SQLite database with the pre-0099 schema,\n seeds test data, runs `_migrate_sqlite`, and verifies all changes.\n \"\"\"\n\n @classmethod\n def setUpClass(cls):\n super().setUpClass()\n cls.conn, cls.hashes = _build_test_db()\n _run_migration_on(cls.conn)\n\n @classmethod\n def tearDownClass(cls):\n cls.conn.close()\n super().tearDownClass()\n\n def _cursor(self):\n return self.conn.cursor()\n\n # -- schema assertions -------------------------------------------------\n\n def test_photo_has_id_column(self):\n cur = self._cursor()\n cols = _sqlite_column_names(cur, \"api_photo\")\n self.assertIn(\"id\", cols)\n\n def test_photo_has_image_hash_column(self):\n cur = self._cursor()\n cols = _sqlite_column_names(cur, \"api_photo\")\n self.assertIn(\"image_hash\", cols)\n\n def test_photo_pk_is_id(self):\n cur = self._cursor()\n pk = _sqlite_pk_columns(cur, \"api_photo\")\n self.assertEqual(pk, [\"id\"])\n\n def test_image_hash_unique_index(self):\n cur = self._cursor()\n self.assertTrue(_sqlite_index_exists(cur, \"api_photo_image_hash_unique\"))\n\n def test_performance_indexes(self):\n expected = [\n \"api_face_photo_id_idx\",\n \"api_photo_shared_to_photo_id_idx\",\n \"api_photo_files_photo_id_idx\",\n \"api_person_cover_photo_id_idx\",\n \"api_albumuser_cover_photo_id_idx\",\n \"api_photostack_primary_photo_id_idx\",\n ]\n cur = self._cursor()\n for idx in expected:\n self.assertTrue(\n _sqlite_index_exists(cur, idx),\n f\"Missing index: {idx}\",\n )\n\n # -- data assertions ---------------------------------------------------\n\n def test_all_photos_have_valid_uuids(self):\n cur = self._cursor()\n cur.execute('SELECT \"id\" FROM api_photo')\n rows = cur.fetchall()\n self.assertEqual(len(rows), 3)\n for (photo_id,) in rows:\n uuid.UUID(photo_id) # will raise if invalid\n\n def test_image_hashes_preserved(self):\n cur = self._cursor()\n cur.execute('SELECT \"image_hash\" FROM api_photo ORDER BY image_hash')\n actual = [r[0] for r in cur.fetchall()]\n self.assertEqual(actual, sorted(self.hashes))\n\n def test_each_photo_has_distinct_uuid(self):\n cur = self._cursor()\n cur.execute('SELECT \"id\" FROM api_photo')\n ids = [r[0] for r in cur.fetchall()]\n self.assertEqual(len(ids), len(set(ids)))\n\n def test_face_fk_translated(self):\n \"\"\"Faces should join to photos via the new UUID id.\"\"\"\n cur = self._cursor()\n cur.execute(\n 'SELECT f.photo_id, p.id FROM api_face f '\n 'JOIN api_photo p ON f.photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 2)\n for fk, pk in rows:\n self.assertEqual(fk, pk)\n uuid.UUID(fk)\n\n def test_no_orphan_faces(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT COUNT(*) FROM api_face f '\n 'LEFT JOIN api_photo p ON f.photo_id = p.id '\n 'WHERE p.id IS NULL'\n )\n self.assertEqual(cur.fetchone()[0], 0)\n\n def test_thumbnail_fk_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT t.photo_id, p.id FROM api_thumbnail t '\n 'JOIN api_photo p ON t.photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 1)\n self.assertEqual(rows[0][0], rows[0][1])\n\n def test_photo_caption_fk_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT c.photo_id, p.id FROM api_photo_caption c '\n 'JOIN api_photo p ON c.photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 1)\n\n def test_photo_search_fk_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT s.photo_id, p.id FROM api_photo_search s '\n 'JOIN api_photo p ON s.photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 1)\n\n def test_person_cover_photo_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT per.cover_photo_id, p.id FROM api_person per '\n 'JOIN api_photo p ON per.cover_photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 1)\n uuid.UUID(rows[0][0])\n\n def test_albumuser_cover_photo_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.cover_photo_id, p.id FROM api_albumuser a '\n 'JOIN api_photo p ON a.cover_photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 1)\n\n def test_m2m_shared_to_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT s.photo_id, p.id FROM api_photo_shared_to s '\n 'JOIN api_photo p ON s.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_m2m_albumuser_photos_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.photo_id, p.id FROM api_albumuser_photos a '\n 'JOIN api_photo p ON a.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_m2m_albumthing_photos_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.photo_id, p.id FROM api_albumthing_photos a '\n 'JOIN api_photo p ON a.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_m2m_albumplace_photos_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.photo_id, p.id FROM api_albumplace_photos a '\n 'JOIN api_photo p ON a.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_m2m_albumdate_photos_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.photo_id, p.id FROM api_albumdate_photos a '\n 'JOIN api_photo p ON a.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_m2m_albumauto_photos_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.photo_id, p.id FROM api_albumauto_photos a '\n 'JOIN api_photo p ON a.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_albumthing_cover_photos_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT a.photo_id, p.id FROM api_albumthing_cover_photos a '\n 'JOIN api_photo p ON a.photo_id = p.id'\n )\n self.assertEqual(len(cur.fetchall()), 1)\n\n def test_photostack_primary_photo_translated(self):\n cur = self._cursor()\n cur.execute(\n 'SELECT s.primary_photo_id, p.id FROM api_photostack s '\n 'JOIN api_photo p ON s.primary_photo_id = p.id'\n )\n rows = cur.fetchall()\n self.assertEqual(len(rows), 1)\n uuid.UUID(rows[0][0])\n\n\n# ============================================================================\n# Test: Post-migration schema on the live Django test database\n# ============================================================================\n\nclass TestPostMigrationSchema(TestCase):\n \"\"\"\n Verify the Django test DB (where migration 0099 already ran during\n test database creation) has the expected post-migration schema.\n\n This validates that the migration ran successfully on whatever backend\n the test suite is configured with (SQLite via test_sqlite settings).\n \"\"\"\n\n def test_photo_table_has_id_and_image_hash(self):\n with connection.cursor() as cursor:\n if connection.vendor == \"sqlite\":\n cursor.execute('PRAGMA table_info(\"api_photo\")')\n cols = {row[1] for row in cursor.fetchall()}\n else:\n cursor.execute(\n \"SELECT column_name FROM information_schema.columns \"\n \"WHERE table_name = 'api_photo'\"\n )\n cols = {row[0] for row in cursor.fetchall()}\n self.assertIn(\"id\", cols)\n self.assertIn(\"image_hash\", cols)\n\n def test_photo_pk_is_uuid_field(self):\n \"\"\"Verify Django's ORM sees the PK as a UUID field named 'id'.\"\"\"\n from api.models import Photo\n pk_field = Photo._meta.pk\n self.assertEqual(pk_field.name, \"id\")\n self.assertIsInstance(pk_field, __import__(\"django\").db.models.UUIDField)\n\n\n# ============================================================================\n# Test: Dispatch and reverse logic\n# ============================================================================\n\nclass TestMigrationDispatch(TestCase):\n \"\"\"Test migrate_forward dispatch and migrate_reverse error.\"\"\"\n\n def test_dispatches_to_sqlite(self):\n mock_editor = MagicMock()\n mock_editor.connection.vendor = \"sqlite\"\n with patch.object(_mod, \"_migrate_sqlite\") as mock_fn:\n _mod.migrate_forward(MagicMock(), mock_editor)\n mock_fn.assert_called_once_with(mock_editor)\n\n def test_dispatches_to_postgresql(self):\n mock_editor = MagicMock()\n mock_editor.connection.vendor = \"postgresql\"\n with patch.object(_mod, \"_migrate_postgresql\") as mock_fn:\n _mod.migrate_forward(MagicMock(), mock_editor)\n mock_fn.assert_called_once_with(mock_editor)\n\n def test_rejects_unknown_backend(self):\n mock_editor = MagicMock()\n mock_editor.connection.vendor = \"oracle\"\n with self.assertRaises(ValueError):\n _mod.migrate_forward(MagicMock(), mock_editor)\n\n def test_reverse_raises_runtime_error(self):\n with self.assertRaises(RuntimeError):\n _mod.migrate_reverse(MagicMock(), MagicMock())\n\n\n# ============================================================================\n# Test: SQLite helper functions in isolation\n# ============================================================================\n\nclass TestSQLiteHelpers(TestCase):\n \"\"\"Unit tests for the individual SQLite helper functions.\"\"\"\n\n def test_column_info(self):\n if connection.vendor != \"sqlite\":\n self.skipTest(\"SQLite-specific\")\n with connection.cursor() as cur:\n cur.execute(\n \"CREATE TABLE IF NOT EXISTS _t_col \"\n \"(pk INTEGER PRIMARY KEY, name TEXT NOT NULL, val REAL)\"\n )\n cols = _mod._sqlite_column_info(cur, \"_t_col\")\n cur.execute(\"DROP TABLE IF EXISTS _t_col\")\n self.assertEqual({c[1] for c in cols}, {\"pk\", \"name\", \"val\"})\n\n def test_index_info(self):\n if connection.vendor != \"sqlite\":\n self.skipTest(\"SQLite-specific\")\n with connection.cursor() as cur:\n cur.execute(\"CREATE TABLE IF NOT EXISTS _t_idx (a TEXT, b TEXT)\")\n cur.execute(\"CREATE INDEX IF NOT EXISTS _t_idx_a ON _t_idx(a)\")\n idxs = _mod._sqlite_index_info(cur, \"_t_idx\")\n cur.execute(\"DROP TABLE IF EXISTS _t_idx\")\n self.assertIn(\"_t_idx_a\", [i[0] for i in idxs])\n\n def test_recreate_table_changes_pk(self):\n if connection.vendor != \"sqlite\":\n self.skipTest(\"SQLite-specific\")\n with connection.cursor() as cur:\n cur.execute(\n \"CREATE TABLE _t_rec (old_pk TEXT PRIMARY KEY, new_pk TEXT, data TEXT)\"\n )\n cur.execute(\"INSERT INTO _t_rec VALUES ('h1', 'u1', 'a')\")\n cur.execute(\"INSERT INTO _t_rec VALUES ('h2', 'u2', 'b')\")\n\n _mod._sqlite_recreate_table(\n cur, \"_t_rec\", pk_column=\"new_pk\",\n column_overrides={\n \"new_pk\": '\"new_pk\" TEXT NOT NULL',\n \"old_pk\": '\"old_pk\" TEXT NOT NULL UNIQUE',\n },\n )\n pk = _sqlite_pk_columns(cur, \"_t_rec\")\n self.assertEqual(pk, [\"new_pk\"])\n cur.execute(\"SELECT old_pk, new_pk, data FROM _t_rec ORDER BY old_pk\")\n self.assertEqual(cur.fetchall(), [(\"h1\", \"u1\", \"a\"), (\"h2\", \"u2\", \"b\")])\n cur.execute(\"DROP TABLE _t_rec\")\n\n def test_update_fk_table_translates_values(self):\n if connection.vendor != \"sqlite\":\n self.skipTest(\"SQLite-specific\")\n with connection.cursor() as cur:\n cur.execute(\"CREATE TABLE _t_parent (id TEXT PRIMARY KEY)\")\n cur.execute(\"INSERT INTO _t_parent VALUES ('uuid-a')\")\n cur.execute(\n \"CREATE TABLE _t_child (id INTEGER PRIMARY KEY, fk TEXT, info TEXT)\"\n )\n cur.execute(\"INSERT INTO _t_child (fk, info) VALUES ('old', 'r1')\")\n cur.execute(\"INSERT INTO _t_child (fk, info) VALUES ('old', 'r2')\")\n\n _mod._sqlite_update_fk_table(cur, \"_t_child\", \"fk\", {\"old\": \"uuid-a\"})\n\n cur.execute(\"SELECT fk FROM _t_child\")\n self.assertTrue(all(r[0] == \"uuid-a\" for r in cur.fetchall()))\n cur.execute(\"DROP TABLE _t_child\")\n cur.execute(\"DROP TABLE _t_parent\")\n\n def test_update_fk_table_skips_missing_table(self):\n if connection.vendor != \"sqlite\":\n self.skipTest(\"SQLite-specific\")\n with connection.cursor() as cur:\n # Should not raise\n _mod._sqlite_update_fk_table(cur, \"_no_such_table_999\", \"col\", {\"a\": \"b\"})\n" }, { "path": "api/tests/test_migration_0101.py", "content": "\"\"\"\nTest for migration 0101_populate_photo_metadata to ensure it works correctly.\n\nThis test verifies:\n1. The fix for the PostgreSQL error: \"operator does not exist: uuid = character varying\"\n2. The fix for the SQLite issue: cursor-during-writes conflict with iterator() + writes\n\"\"\"\nfrom importlib import import_module\n\nfrom django.db import models\nfrom django.db.models import Exists, OuterRef, Subquery\nfrom django.test import TestCase\n\nfrom api.models import Photo\nfrom api.models.photo_caption import PhotoCaption\nfrom api.models.photo_metadata import PhotoMetadata\nfrom api.tests.utils import create_test_photo, create_test_user\n\n_migration = import_module(\"api.migrations.0101_populate_photo_metadata\")\nBATCH_SIZE = _migration.BATCH_SIZE\n\n\nclass Migration0101TestCase(TestCase):\n \"\"\"Test the migration logic for populating PhotoMetadata.\"\"\"\n\n def setUp(self):\n \"\"\"Set up test data.\"\"\"\n self.user = create_test_user()\n\n def test_subquery_with_uuid_primary_key(self):\n \"\"\"\n Test that the subquery correctly references the UUID primary key.\n \n This test ensures the fix for the PostgreSQL error where\n photo_id (UUID) was incorrectly compared with image_hash (VARCHAR).\n \"\"\"\n # Create a photo with caption\n photo = create_test_photo(\n owner=self.user,\n captions_json={\"user_caption\": \"Test caption\", \"keywords\": [\"test\"]},\n )\n\n # Verify the caption was created\n self.assertTrue(PhotoCaption.objects.filter(photo=photo).exists())\n\n # Test the subquery pattern from the migration (FIXED version)\n caption_subquery = PhotoCaption.objects.filter(\n photo_id=OuterRef('pk') # Using 'pk' (UUID) instead of 'image_hash' (VARCHAR)\n ).values('captions_json')[:1]\n\n # Query photos with the subquery annotation\n photos = Photo.objects.annotate(\n captions_data=Subquery(caption_subquery)\n ).filter(pk=photo.pk)\n\n # Verify the query works without PostgreSQL type errors\n self.assertEqual(photos.count(), 1)\n photo_with_caption = photos.first()\n self.assertIsNotNone(photo_with_caption.captions_data)\n self.assertEqual(\n photo_with_caption.captions_data.get(\"user_caption\"),\n \"Test caption\"\n )\n\n def test_migration_logic_creates_metadata(self):\n \"\"\"\n Test the full migration logic to ensure PhotoMetadata is populated.\n \"\"\"\n # Create photos without metadata\n photo1 = create_test_photo(\n owner=self.user,\n captions_json={\"user_caption\": \"First photo\"},\n )\n photo2 = create_test_photo(\n owner=self.user,\n captions_json={\"user_caption\": \"Second photo\", \"keywords\": [\"tag1\", \"tag2\"]},\n )\n\n # Manually delete any metadata that might have been auto-created\n PhotoMetadata.objects.filter(photo__in=[photo1, photo2]).delete()\n\n # Verify no metadata exists\n self.assertFalse(PhotoMetadata.objects.filter(photo=photo1).exists())\n self.assertFalse(PhotoMetadata.objects.filter(photo=photo2).exists())\n\n # Simulate the migration logic\n caption_subquery = PhotoCaption.objects.filter(\n photo_id=OuterRef('pk')\n ).values('captions_json')[:1]\n\n existing_metadata = PhotoMetadata.objects.filter(photo_id=OuterRef('pk'))\n\n photos = Photo.objects.filter(\n ~models.Exists(existing_metadata)\n ).annotate(\n captions_data=Subquery(caption_subquery)\n )\n\n # Create PhotoMetadata for each photo\n for photo in photos:\n captions_json = photo.captions_data\n PhotoMetadata.objects.create(\n photo=photo,\n caption=captions_json.get(\"user_caption\") if captions_json else None,\n keywords=list(captions_json.get(\"keywords\", [])) if captions_json else [],\n source=\"embedded\",\n version=1,\n )\n\n # Verify metadata was created\n self.assertTrue(PhotoMetadata.objects.filter(photo=photo1).exists())\n self.assertTrue(PhotoMetadata.objects.filter(photo=photo2).exists())\n\n # Verify caption data was correctly populated\n metadata1 = PhotoMetadata.objects.get(photo=photo1)\n self.assertEqual(metadata1.caption, \"First photo\")\n\n metadata2 = PhotoMetadata.objects.get(photo=photo2)\n self.assertEqual(metadata2.caption, \"Second photo\")\n self.assertEqual(metadata2.keywords, [\"tag1\", \"tag2\"])\n\n def test_batch_processing_without_iterator(self):\n \"\"\"\n Test the fixed migration approach: fetch IDs upfront, process in batches.\n\n This tests the SQLite-compatible pattern where:\n - All photo IDs are collected first (no open cursor during writes)\n - Caption data is loaded per-batch\n - No iterator() is used alongside writes\n - Photos with pre-existing metadata are correctly excluded\n \"\"\"\n photo1 = create_test_photo(\n owner=self.user,\n captions_json={\"user_caption\": \"Batch photo 1\", \"keywords\": [\"a\"]},\n )\n photo2 = create_test_photo(\n owner=self.user,\n captions_json={\"user_caption\": \"Batch photo 2\", \"keywords\": [\"b\", \"c\"]},\n )\n photo3 = create_test_photo(owner=self.user) # no caption\n\n # photo_already_done has pre-existing metadata — should be excluded\n photo_already_done = create_test_photo(owner=self.user)\n PhotoMetadata.objects.filter(photo=photo_already_done).delete()\n existing_meta = PhotoMetadata.objects.create(\n photo=photo_already_done,\n caption=\"pre-existing\",\n source=\"embedded\",\n version=1,\n )\n\n PhotoMetadata.objects.filter(photo__in=[photo1, photo2, photo3]).delete()\n\n # --- Simulate the fixed migration approach ---\n existing_metadata = PhotoMetadata.objects.filter(photo_id=OuterRef('pk'))\n photo_ids = list(\n Photo.objects\n .filter(~Exists(existing_metadata))\n .values_list('pk', flat=True)\n )\n\n all_batch = []\n for chunk_start in range(0, len(photo_ids), BATCH_SIZE):\n chunk_ids = photo_ids[chunk_start:chunk_start + BATCH_SIZE]\n captions = {\n c.photo_id: c.captions_json\n for c in PhotoCaption.objects.filter(photo_id__in=chunk_ids)\n }\n batch = []\n for photo in Photo.objects.filter(pk__in=chunk_ids):\n captions_json = captions.get(photo.pk)\n batch.append(PhotoMetadata(\n photo=photo,\n caption=captions_json.get(\"user_caption\") if captions_json else None,\n keywords=list(captions_json.get(\"keywords\", [])) if captions_json else [],\n source=\"embedded\",\n version=1,\n ))\n PhotoMetadata.objects.bulk_create(batch, ignore_conflicts=True)\n all_batch.extend(batch)\n\n # All three photos must have metadata\n self.assertTrue(PhotoMetadata.objects.filter(photo=photo1).exists())\n self.assertTrue(PhotoMetadata.objects.filter(photo=photo2).exists())\n self.assertTrue(PhotoMetadata.objects.filter(photo=photo3).exists())\n\n m1 = PhotoMetadata.objects.get(photo=photo1)\n self.assertEqual(m1.caption, \"Batch photo 1\")\n self.assertEqual(m1.keywords, [\"a\"])\n\n m2 = PhotoMetadata.objects.get(photo=photo2)\n self.assertEqual(m2.caption, \"Batch photo 2\")\n self.assertEqual(m2.keywords, [\"b\", \"c\"])\n\n m3 = PhotoMetadata.objects.get(photo=photo3)\n self.assertIsNone(m3.caption)\n self.assertEqual(m3.keywords, [])\n\n # photo_already_done must NOT have been re-processed — exactly one record,\n # the pre-existing one, and its caption must still be the original value.\n self.assertEqual(PhotoMetadata.objects.filter(photo=photo_already_done).count(), 1)\n existing_meta.refresh_from_db()\n self.assertEqual(existing_meta.caption, \"pre-existing\")\n\n def test_batch_processing_is_idempotent(self):\n \"\"\"\n Running the batch-based migration approach twice should not duplicate records.\n \"\"\"\n photo = create_test_photo(\n owner=self.user,\n captions_json={\"user_caption\": \"Idempotent test\"},\n )\n PhotoMetadata.objects.filter(photo=photo).delete()\n\n def run_migration_logic():\n existing_metadata = PhotoMetadata.objects.filter(photo_id=OuterRef('pk'))\n photo_ids = list(\n Photo.objects\n .filter(~Exists(existing_metadata))\n .values_list('pk', flat=True)\n )\n captions = {\n c.photo_id: c.captions_json\n for c in PhotoCaption.objects.filter(photo_id__in=photo_ids)\n }\n batch = []\n for p in Photo.objects.filter(pk__in=photo_ids):\n captions_json = captions.get(p.pk)\n batch.append(PhotoMetadata(\n photo=p,\n caption=captions_json.get(\"user_caption\") if captions_json else None,\n keywords=list(captions_json.get(\"keywords\", [])) if captions_json else [],\n source=\"embedded\",\n version=1,\n ))\n PhotoMetadata.objects.bulk_create(batch, ignore_conflicts=True)\n\n run_migration_logic()\n self.assertEqual(PhotoMetadata.objects.filter(photo=photo).count(), 1)\n\n # Second run: the photo already has metadata, so it should be skipped\n run_migration_logic()\n self.assertEqual(PhotoMetadata.objects.filter(photo=photo).count(), 1)\n\n def test_bulk_create_ignore_conflicts_on_duplicate(self):\n \"\"\"\n Verify that bulk_create(ignore_conflicts=True) silently skips duplicate records.\n\n This tests the safety net used in each batch: if for any reason a\n PhotoMetadata record already exists for a photo in the batch (e.g. a\n partially-applied migration is retried), the insert is ignored rather\n than raising an IntegrityError.\n \"\"\"\n photo = create_test_photo(owner=self.user)\n PhotoMetadata.objects.filter(photo=photo).delete()\n\n first = PhotoMetadata(photo=photo, source=\"embedded\", version=1)\n PhotoMetadata.objects.bulk_create([first], ignore_conflicts=True)\n self.assertEqual(PhotoMetadata.objects.filter(photo=photo).count(), 1)\n\n # Attempt to insert the same photo again — must not raise\n duplicate = PhotoMetadata(photo=photo, source=\"embedded\", version=1)\n PhotoMetadata.objects.bulk_create([duplicate], ignore_conflicts=True)\n self.assertEqual(PhotoMetadata.objects.filter(photo=photo).count(), 1)\n" }, { "path": "api/tests/test_multi_user_isolation.py", "content": "\"\"\"\nTests for multi-user isolation and security.\n\nEnsures that:\n- Users can only see/modify their own duplicates and stacks\n- Shared photos don't leak into other users' detection\n- Admin access is properly scoped\n- Cross-user operations are blocked\n\"\"\"\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient, APITestCase\n\nfrom api.models.duplicate import Duplicate\nfrom api.models.photo_stack import PhotoStack\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass DuplicateUserIsolationTestCase(APITestCase):\n \"\"\"Test that duplicates are properly scoped to users.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.admin = create_test_user()\n self.admin.is_staff = True\n self.admin.save()\n \n self.client = APIClient()\n\n def test_user_cannot_see_other_user_duplicates(self):\n \"\"\"Test that users can only see their own duplicates.\"\"\"\n # Create duplicate for user2\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user2,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n # Login as user1\n self.client.force_authenticate(user=self.user1)\n \n # Try to access duplicate list\n response = self.client.get(\"/api/duplicates\")\n self.assertEqual(response.status_code, 200)\n \n # Should not see user2's duplicate\n dup_ids = [d[\"id\"] for d in response.data.get(\"results\", [])]\n self.assertNotIn(str(dup.id), dup_ids)\n\n def test_user_cannot_access_other_user_duplicate_detail(self):\n \"\"\"Test that users cannot access other users' duplicate details.\"\"\"\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user2,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.get(f\"/api/duplicates/{dup.id}\")\n self.assertIn(response.status_code, [403, 404])\n\n def test_user_cannot_resolve_other_user_duplicate(self):\n \"\"\"Test that users cannot resolve other users' duplicates.\"\"\"\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user2,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n {\"kept_photo_id\": str(photos[0].pk)},\n format=\"json\"\n )\n self.assertIn(response.status_code, [403, 404])\n\n def test_user_cannot_delete_other_user_duplicate(self):\n \"\"\"Test that users cannot delete other users' duplicates.\"\"\"\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user2,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.delete(f\"/api/duplicates/{dup.id}/delete\")\n self.assertIn(response.status_code, [403, 404])\n \n # Duplicate should still exist\n self.assertTrue(Duplicate.objects.filter(pk=dup.pk).exists())\n\n def test_admin_can_see_duplicate_stats(self):\n \"\"\"Test that admin can see global stats.\"\"\"\n # Create duplicates for different users\n for user in [self.user1, self.user2]:\n photos = [create_test_photo(owner=user) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n self.client.force_authenticate(user=self.admin)\n \n response = self.client.get(\"/api/duplicates/stats\")\n self.assertEqual(response.status_code, 200)\n\n\nclass StackUserIsolationTestCase(APITestCase):\n \"\"\"Test that stacks are properly scoped to users.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client = APIClient()\n\n def test_user_cannot_see_other_user_stacks(self):\n \"\"\"Test that users can only see their own stacks.\"\"\"\n # Create stack for user2\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n stack = PhotoStack.objects.create(\n owner=self.user2,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.get(\"/api/stacks\")\n self.assertEqual(response.status_code, 200)\n \n stack_ids = [s[\"id\"] for s in response.data.get(\"results\", [])]\n self.assertNotIn(str(stack.id), stack_ids)\n\n def test_user_cannot_access_other_user_stack_detail(self):\n \"\"\"Test that users cannot access other users' stack details.\"\"\"\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n stack = PhotoStack.objects.create(\n owner=self.user2,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.get(f\"/api/stacks/{stack.id}\")\n self.assertIn(response.status_code, [403, 404])\n\n def test_user_cannot_modify_other_user_stack(self):\n \"\"\"Test that users cannot modify other users' stacks.\"\"\"\n photos = [create_test_photo(owner=self.user2) for _ in range(3)]\n stack = PhotoStack.objects.create(\n owner=self.user2,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos[:2])\n \n self.client.force_authenticate(user=self.user1)\n \n # Try to add a photo\n response = self.client.post(\n f\"/api/stacks/{stack.id}/add\",\n {\"photo_ids\": [str(photos[2].pk)]},\n format=\"json\"\n )\n self.assertIn(response.status_code, [403, 404])\n\n def test_user_cannot_delete_other_user_stack(self):\n \"\"\"Test that users cannot delete other users' stacks.\"\"\"\n photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n stack = PhotoStack.objects.create(\n owner=self.user2,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.delete(f\"/api/stacks/{stack.id}/delete\")\n self.assertIn(response.status_code, [403, 404])\n \n # Stack should still exist\n self.assertTrue(PhotoStack.objects.filter(pk=stack.pk).exists())\n\n def test_user_cannot_create_stack_with_other_user_photos(self):\n \"\"\"Test that users cannot create stacks with other users' photos.\"\"\"\n other_photos = [create_test_photo(owner=self.user2) for _ in range(2)]\n \n self.client.force_authenticate(user=self.user1)\n \n response = self.client.post(\n \"/api/stacks/manual\",\n {\"photo_ids\": [str(p.pk) for p in other_photos]},\n format=\"json\"\n )\n # Should either fail or create empty stack\n if response.status_code == 201:\n # If created, should have no photos\n stack_id = response.data.get(\"id\")\n if stack_id:\n stack = PhotoStack.objects.get(pk=stack_id)\n self.assertEqual(stack.photos.count(), 0)\n else:\n self.assertIn(response.status_code, [400, 403, 404])\n\n\nclass SharedPhotoIsolationTestCase(TestCase):\n \"\"\"Test that shared photos don't affect personal stacks/duplicates.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n\n def test_shared_photo_not_in_receiver_stacks(self):\n \"\"\"Test that photos shared TO a user don't appear in their stack detection.\"\"\"\n # Create photo for user1\n photo1 = create_test_photo(owner=self.user1)\n \n # Share with user2\n photo1.shared_to.add(self.user2)\n \n # Create a stack for user1\n photo2 = create_test_photo(owner=self.user1)\n stack = PhotoStack.objects.create(\n owner=self.user1,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(photo1, photo2)\n \n # User2 should not see this stack\n user2_stacks = PhotoStack.objects.filter(owner=self.user2)\n self.assertEqual(user2_stacks.count(), 0)\n\n def test_shared_photo_not_in_receiver_duplicates(self):\n \"\"\"Test that shared photos don't appear in receiver's duplicate detection.\"\"\"\n # Create photos for user1\n photos = [create_test_photo(owner=self.user1) for _ in range(2)]\n \n # Create duplicate group for user1\n dup = Duplicate.objects.create(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n # Share one photo with user2\n photos[0].shared_to.add(self.user2)\n \n # User2 should not see this duplicate\n user2_dups = Duplicate.objects.filter(owner=self.user2)\n self.assertEqual(user2_dups.count(), 0)\n\n def test_user_stack_unaffected_by_shared_photos(self):\n \"\"\"Test that user's own stacks aren't affected by sharing.\"\"\"\n # Create stack for user1\n photos = [create_test_photo(owner=self.user1) for _ in range(3)]\n stack = PhotoStack.objects.create(\n owner=self.user1,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(*photos)\n \n # Share one photo\n photos[0].shared_to.add(self.user2)\n \n # Stack should still have all 3 photos\n stack.refresh_from_db()\n self.assertEqual(stack.photos.count(), 3)\n\n\nclass DetectionUserIsolationTestCase(APITestCase):\n \"\"\"Test that detection operations are properly isolated.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client = APIClient()\n\n def test_duplicate_detection_only_affects_own_photos(self):\n \"\"\"Test that duplicate detection only processes user's own photos.\"\"\"\n # Create photos for both users with same hash (simulating duplicates)\n photo1_u1 = create_test_photo(owner=self.user1)\n photo2_u1 = create_test_photo(owner=self.user1)\n photo1_u2 = create_test_photo(owner=self.user2)\n photo2_u2 = create_test_photo(owner=self.user2)\n \n # Set same perceptual hash to simulate visual duplicates\n same_hash = \"0\" * 16\n for photo in [photo1_u1, photo2_u1, photo1_u2, photo2_u2]:\n photo.image_phash = same_hash\n photo.save()\n \n self.client.force_authenticate(user=self.user1)\n \n # Trigger detection for user1\n response = self.client.post(\"/api/duplicates/detect\")\n self.assertIn(response.status_code, [200, 202])\n \n # User2's photos should not be in user1's duplicates\n user1_dups = Duplicate.objects.filter(owner=self.user1)\n for dup in user1_dups:\n for photo in dup.photos.all():\n self.assertEqual(photo.owner, self.user1)\n\n def test_stack_detection_only_affects_own_photos(self):\n \"\"\"Test that stack detection only processes user's own photos.\"\"\"\n # Create photos for both users\n for _ in range(3):\n create_test_photo(owner=self.user1)\n create_test_photo(owner=self.user2)\n \n self.client.force_authenticate(user=self.user1)\n \n # Trigger stack detection for user1\n response = self.client.post(\"/api/stacks/detect\")\n self.assertIn(response.status_code, [200, 202])\n \n # User2 should have no stacks created\n user2_stacks = PhotoStack.objects.filter(owner=self.user2)\n self.assertEqual(user2_stacks.count(), 0)\n\n\nclass CrossUserOperationTestCase(APITestCase):\n \"\"\"Test that cross-user operations are properly blocked.\"\"\"\n\n def setUp(self):\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client = APIClient()\n\n def test_cannot_add_other_user_photo_to_own_stack(self):\n \"\"\"Test that users cannot add other users' photos to their stacks.\"\"\"\n own_photos = [create_test_photo(owner=self.user1) for _ in range(2)]\n other_photo = create_test_photo(owner=self.user2)\n \n stack = PhotoStack.objects.create(\n owner=self.user1,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*own_photos)\n \n self.client.force_authenticate(user=self.user1)\n \n _response = self.client.post(\n f\"/api/stacks/{stack.id}/add\",\n {\"photo_ids\": [str(other_photo.pk)]},\n format=\"json\"\n )\n \n # Should either fail or not add the photo\n stack.refresh_from_db()\n self.assertNotIn(other_photo, stack.photos.all())\n\n def test_cannot_resolve_duplicate_with_other_user_photo(self):\n \"\"\"Test that users cannot resolve duplicates by keeping other user's photo.\"\"\"\n photos = [create_test_photo(owner=self.user1) for _ in range(2)]\n other_photo = create_test_photo(owner=self.user2)\n \n dup = Duplicate.objects.create(\n owner=self.user1,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n self.client.force_authenticate(user=self.user1)\n \n # Try to resolve keeping other user's photo\n response = self.client.post(\n f\"/api/duplicates/{dup.id}/resolve\",\n {\"kept_photo_id\": str(other_photo.pk)},\n format=\"json\"\n )\n \n # Should fail\n self.assertIn(response.status_code, [400, 403, 404])\n\n def test_cannot_set_other_user_photo_as_stack_primary(self):\n \"\"\"Test that users cannot set other user's photo as stack primary.\"\"\"\n own_photos = [create_test_photo(owner=self.user1) for _ in range(2)]\n other_photo = create_test_photo(owner=self.user2)\n \n stack = PhotoStack.objects.create(\n owner=self.user1,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack.photos.add(*own_photos)\n \n self.client.force_authenticate(user=self.user1)\n \n _response = self.client.post(\n f\"/api/stacks/{stack.id}/primary\",\n {\"photo_id\": str(other_photo.pk)},\n format=\"json\"\n )\n \n # Should fail or not set the photo\n stack.refresh_from_db()\n self.assertNotEqual(stack.primary_photo, other_photo)\n\n\nclass AdminAccessTestCase(APITestCase):\n \"\"\"Test admin access and capabilities.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.admin = create_test_user()\n self.admin.is_staff = True\n self.admin.save()\n self.client = APIClient()\n\n def test_admin_can_view_stats_for_all_users(self):\n \"\"\"Test that admin can view aggregate stats.\"\"\"\n # Create data for regular user\n photos = [create_test_photo(owner=self.user) for _ in range(2)]\n dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n dup.photos.add(*photos)\n \n self.client.force_authenticate(user=self.admin)\n \n response = self.client.get(\"/api/duplicates/stats\")\n self.assertEqual(response.status_code, 200)\n\n def test_regular_user_sees_only_own_stats(self):\n \"\"\"Test that regular users only see their own stats.\"\"\"\n # Create data for admin\n admin_photos = [create_test_photo(owner=self.admin) for _ in range(2)]\n admin_dup = Duplicate.objects.create(\n owner=self.admin,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n admin_dup.photos.add(*admin_photos)\n \n # Create data for user\n user_photos = [create_test_photo(owner=self.user) for _ in range(2)]\n user_dup = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n user_dup.photos.add(*user_photos)\n \n self.client.force_authenticate(user=self.user)\n \n response = self.client.get(\"/api/duplicates/stats\")\n self.assertEqual(response.status_code, 200)\n \n # Stats should reflect only user's data\n # The exact assertion depends on the stats endpoint response format\n" }, { "path": "api/tests/test_only_photos_or_only_videos.py", "content": "from django.test import TestCase\nfrom django.utils import timezone\nfrom rest_framework.test import APIClient\n\nfrom api.models.album_date import AlbumDate\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass OnlyPhotosOrOnlyVideosTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n\n def test_only_photos(self):\n now = timezone.now()\n photo = create_test_photo(owner=self.user, added_on=now, public=True)\n\n album = AlbumDate(owner=self.user)\n album.id = 1\n album.photos.add(photo)\n album.save()\n\n response = self.client.get(\"/api/albums/date/list?photo=true\").url\n response = self.client.get(response)\n\n data = response.json()\n self.assertEqual(1, len(data[\"results\"]))\n" }, { "path": "api/tests/test_perceptual_hash.py", "content": "\"\"\"\nComprehensive tests for api/perceptual_hash.py\n\nTests the perceptual hashing algorithm used for visual duplicate detection:\n- calculate_perceptual_hash: Calculates pHash from image files\n- calculate_hash_from_thumbnail: Wrapper for thumbnail hashing\n- hamming_distance: Calculates bit difference between hashes\n- are_duplicates: Determines if two images are duplicates based on hash similarity\n- find_similar_hashes: Finds all similar hashes in a list\n\"\"\"\n\nimport os\nimport tempfile\nfrom unittest.mock import patch\n\nfrom django.test import TestCase\nfrom PIL import Image\n\nfrom api.perceptual_hash import (\n DEFAULT_HAMMING_THRESHOLD,\n are_duplicates,\n calculate_hash_from_thumbnail,\n calculate_perceptual_hash,\n find_similar_hashes,\n hamming_distance,\n)\n\n\nclass HammingDistanceTestCase(TestCase):\n \"\"\"Tests for the hamming_distance function.\"\"\"\n\n def test_identical_hashes_return_zero(self):\n \"\"\"Identical hashes should have distance 0.\"\"\"\n hash1 = \"a\" * 16 # 64-bit hash as 16 hex chars\n self.assertEqual(hamming_distance(hash1, hash1), 0)\n\n def test_completely_different_hashes(self):\n \"\"\"Completely different hashes should have maximum distance (64 for 64-bit hash).\"\"\"\n # 0000...0000 vs FFFF...FFFF\n hash1 = \"0\" * 16\n hash2 = \"f\" * 16\n distance = hamming_distance(hash1, hash2)\n self.assertEqual(distance, 64) # All 64 bits different\n\n def test_one_bit_difference(self):\n \"\"\"Hashes differing by one bit should have distance 1.\"\"\"\n # 0x0 = 0000 in binary, 0x1 = 0001 in binary - 1 bit different\n hash1 = \"0\" * 16\n hash2 = \"0\" * 15 + \"1\"\n distance = hamming_distance(hash1, hash2)\n self.assertEqual(distance, 1)\n\n def test_half_bits_different(self):\n \"\"\"Test hashes with approximately half the bits different.\"\"\"\n # 0 = 0000, a = 1010 - 2 bits different per hex char\n hash1 = \"0\" * 16\n hash2 = \"a\" * 16 # 2 bits per char * 16 chars = 32 bits\n distance = hamming_distance(hash1, hash2)\n self.assertEqual(distance, 32)\n\n def test_invalid_hash_returns_max_distance(self):\n \"\"\"Invalid hash strings should return maximum distance (64).\"\"\"\n distance = hamming_distance(\"invalid\", \"hash\")\n self.assertEqual(distance, 64)\n\n def test_empty_strings_return_max_distance(self):\n \"\"\"Empty strings should return maximum distance.\"\"\"\n distance = hamming_distance(\"\", \"\")\n self.assertEqual(distance, 64)\n\n def test_mixed_valid_invalid_returns_max_distance(self):\n \"\"\"Mix of valid and invalid should return max distance.\"\"\"\n valid_hash = \"a\" * 16\n distance = hamming_distance(valid_hash, \"not_hex_xyz\")\n self.assertEqual(distance, 64)\n\n def test_different_length_hashes(self):\n \"\"\"Different length hashes should return max distance or handle gracefully.\"\"\"\n hash1 = \"a\" * 16\n hash2 = \"a\" * 8 # Shorter hash\n distance = hamming_distance(hash1, hash2)\n # imagehash may handle this differently - should not crash\n self.assertIsInstance(distance, int)\n\n def test_real_phash_values(self):\n \"\"\"Test with realistic pHash values.\"\"\"\n # These are example pHash values that might be generated\n hash1 = \"8f94b5a16363c3c3\"\n hash2 = \"8f94b5a16363c3c7\" # 2 bits different (c3 vs c7)\n distance = hamming_distance(hash1, hash2)\n self.assertLessEqual(distance, 5) # Should be small\n\n def test_case_insensitive_hashes(self):\n \"\"\"Hash comparison should be case-insensitive (hex).\"\"\"\n hash1 = \"ABCDEF0123456789\"\n hash2 = \"abcdef0123456789\"\n distance = hamming_distance(hash1, hash2)\n self.assertEqual(distance, 0)\n\n\nclass AreDuplicatesTestCase(TestCase):\n \"\"\"Tests for the are_duplicates function.\"\"\"\n\n def test_identical_hashes_are_duplicates(self):\n \"\"\"Identical hashes should always be considered duplicates.\"\"\"\n hash1 = \"a\" * 16\n self.assertTrue(are_duplicates(hash1, hash1))\n\n def test_distance_under_threshold_is_duplicate(self):\n \"\"\"Hashes with distance under threshold are duplicates.\"\"\"\n # Using 1 bit difference which is well under default threshold of 10\n hash1 = \"0\" * 16\n hash2 = \"0\" * 15 + \"1\"\n self.assertTrue(are_duplicates(hash1, hash2))\n\n def test_distance_at_threshold_is_duplicate(self):\n \"\"\"Hashes with distance exactly at threshold are duplicates.\"\"\"\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=10):\n self.assertTrue(are_duplicates(\"a\" * 16, \"b\" * 16, threshold=10))\n\n def test_distance_over_threshold_not_duplicate(self):\n \"\"\"Hashes with distance over threshold are not duplicates.\"\"\"\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=11):\n self.assertFalse(are_duplicates(\"a\" * 16, \"b\" * 16, threshold=10))\n\n def test_empty_hash1_not_duplicate(self):\n \"\"\"Empty first hash should not be considered duplicate.\"\"\"\n self.assertFalse(are_duplicates(\"\", \"a\" * 16))\n\n def test_empty_hash2_not_duplicate(self):\n \"\"\"Empty second hash should not be considered duplicate.\"\"\"\n self.assertFalse(are_duplicates(\"a\" * 16, \"\"))\n\n def test_none_hash1_not_duplicate(self):\n \"\"\"None first hash should not be considered duplicate.\"\"\"\n self.assertFalse(are_duplicates(None, \"a\" * 16))\n\n def test_none_hash2_not_duplicate(self):\n \"\"\"None second hash should not be considered duplicate.\"\"\"\n self.assertFalse(are_duplicates(\"a\" * 16, None))\n\n def test_both_none_not_duplicate(self):\n \"\"\"Both None should not be considered duplicate.\"\"\"\n self.assertFalse(are_duplicates(None, None))\n\n def test_both_empty_not_duplicate(self):\n \"\"\"Both empty should not be considered duplicate.\"\"\"\n self.assertFalse(are_duplicates(\"\", \"\"))\n\n def test_custom_threshold_strict(self):\n \"\"\"Strict threshold (lower) should reject more.\"\"\"\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=5):\n self.assertTrue(are_duplicates(\"a\" * 16, \"b\" * 16, threshold=5))\n self.assertFalse(are_duplicates(\"a\" * 16, \"b\" * 16, threshold=4))\n\n def test_custom_threshold_loose(self):\n \"\"\"Loose threshold (higher) should accept more.\"\"\"\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=15):\n self.assertTrue(are_duplicates(\"a\" * 16, \"b\" * 16, threshold=15))\n self.assertTrue(are_duplicates(\"a\" * 16, \"b\" * 16, threshold=20))\n\n def test_default_threshold_value(self):\n \"\"\"Default threshold should be 10.\"\"\"\n self.assertEqual(DEFAULT_HAMMING_THRESHOLD, 10)\n\n\nclass FindSimilarHashesTestCase(TestCase):\n \"\"\"Tests for the find_similar_hashes function.\"\"\"\n\n def test_empty_target_hash_returns_empty(self):\n \"\"\"Empty target hash should return empty list.\"\"\"\n hash_list = [(\"img1\", \"a\" * 16), (\"img2\", \"b\" * 16)]\n result = find_similar_hashes(\"\", hash_list)\n self.assertEqual(result, [])\n\n def test_none_target_hash_returns_empty(self):\n \"\"\"None target hash should return empty list.\"\"\"\n hash_list = [(\"img1\", \"a\" * 16), (\"img2\", \"b\" * 16)]\n result = find_similar_hashes(None, hash_list)\n self.assertEqual(result, [])\n\n def test_empty_hash_list_returns_empty(self):\n \"\"\"Empty hash list should return empty list.\"\"\"\n result = find_similar_hashes(\"a\" * 16, [])\n self.assertEqual(result, [])\n\n def test_finds_similar_hashes(self):\n \"\"\"Should find hashes within threshold.\"\"\"\n target = \"0\" * 16\n # 1 bit different - should be found\n similar = \"0\" * 15 + \"1\"\n hash_list = [(\"img1\", similar)]\n result = find_similar_hashes(target, hash_list)\n self.assertEqual(len(result), 1)\n self.assertEqual(result[0][0], \"img1\")\n self.assertEqual(result[0][1], 1) # distance of 1\n\n def test_excludes_distant_hashes(self):\n \"\"\"Should exclude hashes beyond threshold.\"\"\"\n target = \"0\" * 16\n distant = \"f\" * 16 # 64 bits different\n hash_list = [(\"img1\", distant)]\n result = find_similar_hashes(target, hash_list)\n self.assertEqual(result, [])\n\n def test_skips_identical_hash(self):\n \"\"\"Should skip exact same hash (self-comparison).\"\"\"\n target = \"a\" * 16\n hash_list = [(\"img1\", target)] # Same hash\n result = find_similar_hashes(target, hash_list)\n self.assertEqual(result, [])\n\n def test_skips_none_hash_in_list(self):\n \"\"\"Should skip None hashes in the list.\"\"\"\n target = \"a\" * 16\n similar = \"a\" * 15 + \"0\" # 1 bit different\n hash_list = [(\"img1\", None), (\"img2\", similar)]\n result = find_similar_hashes(target, hash_list)\n self.assertEqual(len(result), 1)\n self.assertEqual(result[0][0], \"img2\")\n\n def test_skips_empty_hash_in_list(self):\n \"\"\"Should skip empty hashes in the list.\"\"\"\n target = \"a\" * 16\n similar = \"a\" * 15 + \"0\"\n hash_list = [(\"img1\", \"\"), (\"img2\", similar)]\n result = find_similar_hashes(target, hash_list)\n self.assertEqual(len(result), 1)\n self.assertEqual(result[0][0], \"img2\")\n\n def test_sorted_by_distance(self):\n \"\"\"Results should be sorted by distance (closest first).\"\"\"\n target = \"0\" * 16\n # Create hashes with known distances\n with patch(\"api.perceptual_hash.hamming_distance\") as mock_dist:\n mock_dist.side_effect = [8, 3, 5] # Distances for 3 hashes\n hash_list = [(\"img1\", \"h1\"), (\"img2\", \"h2\"), (\"img3\", \"h3\")]\n result = find_similar_hashes(target, hash_list, threshold=10)\n # Should be sorted: img2 (3), img3 (5), img1 (8)\n self.assertEqual(result[0][0], \"img2\")\n self.assertEqual(result[1][0], \"img3\")\n self.assertEqual(result[2][0], \"img1\")\n\n def test_custom_threshold(self):\n \"\"\"Should respect custom threshold.\"\"\"\n target = \"0\" * 16\n with patch(\"api.perceptual_hash.hamming_distance\") as mock_dist:\n mock_dist.return_value = 5\n hash_list = [(\"img1\", \"h1\")]\n # Threshold 4 - should exclude distance 5\n result = find_similar_hashes(target, hash_list, threshold=4)\n self.assertEqual(result, [])\n # Threshold 5 - should include distance 5\n result = find_similar_hashes(target, hash_list, threshold=5)\n self.assertEqual(len(result), 1)\n\n def test_returns_correct_tuple_format(self):\n \"\"\"Results should be (image_id, distance) tuples.\"\"\"\n target = \"0\" * 16\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=2):\n hash_list = [(\"my_image_id\", \"some_hash\")]\n result = find_similar_hashes(target, hash_list)\n self.assertEqual(len(result), 1)\n image_id, distance = result[0]\n self.assertEqual(image_id, \"my_image_id\")\n self.assertEqual(distance, 2)\n\n def test_multiple_similar_all_returned(self):\n \"\"\"All similar hashes should be returned.\"\"\"\n target = \"0\" * 16\n with patch(\"api.perceptual_hash.hamming_distance\", return_value=5):\n hash_list = [(\"img1\", \"h1\"), (\"img2\", \"h2\"), (\"img3\", \"h3\")]\n result = find_similar_hashes(target, hash_list, threshold=10)\n self.assertEqual(len(result), 3)\n\n\nclass CalculatePerceptualHashTestCase(TestCase):\n \"\"\"Tests for the calculate_perceptual_hash function.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test images.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def _create_test_image(self, filename, size=(100, 100), mode=\"RGB\", color=(255, 0, 0)):\n \"\"\"Helper to create a test image file.\"\"\"\n path = os.path.join(self.temp_dir, filename)\n img = Image.new(mode, size, color)\n img.save(path)\n return path\n\n def test_valid_rgb_image(self):\n \"\"\"Should calculate hash for valid RGB image.\"\"\"\n path = self._create_test_image(\"test.jpg\", mode=\"RGB\")\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n self.assertIsInstance(result, str)\n self.assertEqual(len(result), 16) # 64-bit hash = 16 hex chars\n\n def test_valid_rgba_image_converted(self):\n \"\"\"Should handle RGBA images by converting to RGB.\"\"\"\n path = self._create_test_image(\"test.png\", mode=\"RGBA\", color=(255, 0, 0, 128))\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n self.assertEqual(len(result), 16)\n\n def test_valid_grayscale_image(self):\n \"\"\"Should handle grayscale (L mode) images.\"\"\"\n path = self._create_test_image(\"test_gray.jpg\", mode=\"L\", color=128)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n self.assertEqual(len(result), 16)\n\n def test_valid_palette_image_converted(self):\n \"\"\"Should handle palette (P mode) images by converting to RGB.\"\"\"\n path = os.path.join(self.temp_dir, \"test_palette.png\")\n img = Image.new(\"P\", (100, 100))\n img.save(path)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_nonexistent_file_returns_none(self):\n \"\"\"Should return None for nonexistent file.\"\"\"\n result = calculate_perceptual_hash(\"/nonexistent/path/image.jpg\")\n self.assertIsNone(result)\n\n def test_corrupted_file_returns_none(self):\n \"\"\"Should return None for corrupted/invalid image file.\"\"\"\n path = os.path.join(self.temp_dir, \"corrupted.jpg\")\n with open(path, \"w\") as f:\n f.write(\"not an image file content\")\n result = calculate_perceptual_hash(path)\n self.assertIsNone(result)\n\n def test_empty_file_returns_none(self):\n \"\"\"Should return None for empty file.\"\"\"\n path = os.path.join(self.temp_dir, \"empty.jpg\")\n with open(path, \"w\") as _f:\n pass # Create empty file\n result = calculate_perceptual_hash(path)\n self.assertIsNone(result)\n\n def test_directory_instead_of_file_returns_none(self):\n \"\"\"Should return None if path is a directory.\"\"\"\n result = calculate_perceptual_hash(self.temp_dir)\n self.assertIsNone(result)\n\n def test_custom_hash_size(self):\n \"\"\"Should support custom hash sizes.\"\"\"\n path = self._create_test_image(\"test.jpg\")\n # hash_size=16 produces 256-bit hash = 64 hex chars\n result = calculate_perceptual_hash(path, hash_size=16)\n self.assertIsNotNone(result)\n self.assertEqual(len(result), 64)\n\n def test_small_hash_size(self):\n \"\"\"Should support smaller hash sizes.\"\"\"\n path = self._create_test_image(\"test.jpg\")\n # hash_size=4 produces 16-bit hash = 4 hex chars\n result = calculate_perceptual_hash(path, hash_size=4)\n self.assertIsNotNone(result)\n self.assertEqual(len(result), 4)\n\n def test_similar_images_similar_hashes(self):\n \"\"\"Similar images should produce similar hashes.\"\"\"\n # Create two similar images (same color, slight size difference)\n path1 = self._create_test_image(\"img1.jpg\", size=(100, 100), color=(255, 0, 0))\n path2 = self._create_test_image(\"img2.jpg\", size=(110, 110), color=(255, 0, 0))\n hash1 = calculate_perceptual_hash(path1)\n hash2 = calculate_perceptual_hash(path2)\n self.assertIsNotNone(hash1)\n self.assertIsNotNone(hash2)\n # Similar solid color images should have low distance\n distance = hamming_distance(hash1, hash2)\n self.assertLessEqual(distance, 10)\n\n def test_different_images_different_hashes(self):\n \"\"\"Very different images should produce different hashes.\"\"\"\n # Create two very different images with patterns (not solid colors)\n # Solid colors produce similar hashes because pHash uses DCT\n path1 = os.path.join(self.temp_dir, \"pattern1.jpg\")\n path2 = os.path.join(self.temp_dir, \"pattern2.jpg\")\n\n # Create a horizontal gradient pattern\n img1 = Image.new(\"RGB\", (100, 100))\n for x in range(100):\n for y in range(100):\n img1.putpixel((x, y), (x * 2, 0, 0))\n img1.save(path1)\n\n # Create a vertical gradient pattern (different structure)\n img2 = Image.new(\"RGB\", (100, 100))\n for x in range(100):\n for y in range(100):\n img2.putpixel((x, y), (0, 0, y * 2))\n img2.save(path2)\n\n hash1 = calculate_perceptual_hash(path1)\n hash2 = calculate_perceptual_hash(path2)\n self.assertIsNotNone(hash1)\n self.assertIsNotNone(hash2)\n # Different patterns should have noticeable distance\n distance = hamming_distance(hash1, hash2)\n self.assertGreater(distance, 0)\n\n def test_deterministic_hash(self):\n \"\"\"Same image should always produce same hash.\"\"\"\n path = self._create_test_image(\"test.jpg\")\n hash1 = calculate_perceptual_hash(path)\n hash2 = calculate_perceptual_hash(path)\n self.assertEqual(hash1, hash2)\n\n def test_very_small_image(self):\n \"\"\"Should handle very small images (1x1 pixel).\"\"\"\n path = self._create_test_image(\"tiny.jpg\", size=(1, 1))\n _result = calculate_perceptual_hash(path)\n # Should not crash - may return hash or None depending on implementation\n # The key is it doesn't raise an exception\n\n def test_very_large_image(self):\n \"\"\"Should handle large images (though may be slow).\"\"\"\n path = self._create_test_image(\"large.jpg\", size=(1000, 1000))\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n self.assertEqual(len(result), 16)\n\n def test_jpeg_vs_png_same_content(self):\n \"\"\"Same image content in different formats should have similar hash.\"\"\"\n # Create same color image in different formats\n jpg_path = self._create_test_image(\"test.jpg\", color=(100, 150, 200))\n png_path = self._create_test_image(\"test.png\", color=(100, 150, 200))\n hash_jpg = calculate_perceptual_hash(jpg_path)\n hash_png = calculate_perceptual_hash(png_path)\n self.assertIsNotNone(hash_jpg)\n self.assertIsNotNone(hash_png)\n distance = hamming_distance(hash_jpg, hash_png)\n # Same content should have identical or very similar hashes\n self.assertLessEqual(distance, 5)\n\n\nclass CalculateHashFromThumbnailTestCase(TestCase):\n \"\"\"Tests for the calculate_hash_from_thumbnail function.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test images.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_delegates_to_calculate_perceptual_hash(self):\n \"\"\"Should delegate to calculate_perceptual_hash.\"\"\"\n with patch(\"api.perceptual_hash.calculate_perceptual_hash\") as mock:\n mock.return_value = \"abc123\"\n result = calculate_hash_from_thumbnail(\"/some/path\")\n mock.assert_called_once_with(\"/some/path\")\n self.assertEqual(result, \"abc123\")\n\n def test_returns_none_on_failure(self):\n \"\"\"Should return None if underlying function fails.\"\"\"\n result = calculate_hash_from_thumbnail(\"/nonexistent/path\")\n self.assertIsNone(result)\n\n\nclass EdgeCasesTestCase(TestCase):\n \"\"\"Edge case tests for the perceptual hash module.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test images.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_unicode_filename(self):\n \"\"\"Should handle unicode characters in filename.\"\"\"\n path = os.path.join(self.temp_dir, \"图片_照片_🖼️.jpg\")\n img = Image.new(\"RGB\", (50, 50), (255, 255, 0))\n img.save(path)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_special_characters_in_path(self):\n \"\"\"Should handle special characters in file path.\"\"\"\n path = os.path.join(self.temp_dir, \"test with spaces & special (1).jpg\")\n img = Image.new(\"RGB\", (50, 50), (255, 255, 0))\n img.save(path)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_hash_only_contains_hex_chars(self):\n \"\"\"Generated hash should only contain valid hex characters.\"\"\"\n path = os.path.join(self.temp_dir, \"test.jpg\")\n img = Image.new(\"RGB\", (50, 50), (123, 45, 67))\n img.save(path)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n # Check all characters are hex\n valid_hex = set(\"0123456789abcdef\")\n self.assertTrue(all(c in valid_hex for c in result.lower()))\n\n def test_webp_format(self):\n \"\"\"Should handle WebP format images.\"\"\"\n path = os.path.join(self.temp_dir, \"test.webp\")\n img = Image.new(\"RGB\", (50, 50), (100, 100, 100))\n img.save(path, \"WEBP\")\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_gif_format(self):\n \"\"\"Should handle GIF format images.\"\"\"\n path = os.path.join(self.temp_dir, \"test.gif\")\n img = Image.new(\"RGB\", (50, 50), (50, 100, 150))\n img.save(path, \"GIF\")\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_bmp_format(self):\n \"\"\"Should handle BMP format images.\"\"\"\n path = os.path.join(self.temp_dir, \"test.bmp\")\n img = Image.new(\"RGB\", (50, 50), (200, 100, 50))\n img.save(path, \"BMP\")\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_hamming_distance_with_newlines_in_hash(self):\n \"\"\"Should handle hashes that might have whitespace (edge case).\"\"\"\n # This tests robustness - real hashes shouldn't have whitespace\n # imagehash's hex_to_hash is resilient and strips/ignores trailing chars\n distance = hamming_distance(\"a\" * 16 + \"\\n\", \"a\" * 16)\n # The library handles this gracefully - doesn't crash\n self.assertIsInstance(distance, int)\n\n def test_find_similar_with_large_list(self):\n \"\"\"Should handle large hash lists efficiently.\"\"\"\n target = \"0\" * 16\n # Create a large list of hashes\n hash_list = [(f\"img_{i}\", f\"{i:016x}\") for i in range(1000)]\n # Should not crash or hang\n result = find_similar_hashes(target, hash_list, threshold=10)\n self.assertIsInstance(result, list)\n\n def test_are_duplicates_with_whitespace_only_hash(self):\n \"\"\"Should handle whitespace-only hash gracefully.\"\"\"\n self.assertFalse(are_duplicates(\" \", \"a\" * 16))\n self.assertFalse(are_duplicates(\"a\" * 16, \" \"))\n\n def test_cmyk_image_converted(self):\n \"\"\"Should handle CMYK images by converting to RGB.\"\"\"\n path = os.path.join(self.temp_dir, \"test_cmyk.jpg\")\n # Create a CMYK image\n img = Image.new(\"CMYK\", (50, 50), (0, 100, 100, 0)) # Red in CMYK\n img.save(path)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_1bit_image(self):\n \"\"\"Should handle 1-bit (black and white) images.\"\"\"\n path = os.path.join(self.temp_dir, \"test_1bit.png\")\n img = Image.new(\"1\", (50, 50), 1) # White\n img.save(path)\n result = calculate_perceptual_hash(path)\n self.assertIsNotNone(result)\n\n def test_concurrent_hash_calculation(self):\n \"\"\"Hash calculation should be thread-safe (no shared mutable state).\"\"\"\n import concurrent.futures\n\n # Create multiple test images with distinct patterns (not just solid colors)\n paths = []\n for i in range(5):\n path = os.path.join(self.temp_dir, f\"concurrent_{i}.jpg\")\n img = Image.new(\"RGB\", (50, 50))\n # Create distinct patterns for each image\n for x in range(50):\n for y in range(50):\n # Each image has a unique pattern based on i\n if i == 0:\n img.putpixel((x, y), (x * 5, 0, 0)) # Horizontal red gradient\n elif i == 1:\n img.putpixel((x, y), (0, y * 5, 0)) # Vertical green gradient\n elif i == 2:\n img.putpixel((x, y), (0, 0, (x + y) * 2)) # Diagonal blue\n elif i == 3:\n img.putpixel((x, y), ((x * y) % 256, 0, 0)) # Multiplicative pattern\n else:\n img.putpixel((x, y), (255 if x > 25 else 0, 255 if y > 25 else 0, 0)) # Quadrants\n img.save(path)\n paths.append(path)\n\n # Calculate hashes concurrently\n with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:\n futures = [executor.submit(calculate_perceptual_hash, p) for p in paths]\n results = [f.result() for f in futures]\n\n # All should succeed\n self.assertTrue(all(r is not None for r in results))\n # Most should be unique (distinct patterns) - allow some similarity\n unique_count = len(set(results))\n self.assertGreaterEqual(unique_count, 3) # At least 3 unique hashes from 5 distinct patterns\n\n\nclass PerformanceTestCase(TestCase):\n \"\"\"Performance-related tests for the perceptual hash module.\"\"\"\n\n def setUp(self):\n \"\"\"Create temporary directory for test images.\"\"\"\n self.temp_dir = tempfile.mkdtemp()\n\n def tearDown(self):\n \"\"\"Clean up temporary files.\"\"\"\n import shutil\n\n shutil.rmtree(self.temp_dir, ignore_errors=True)\n\n def test_hamming_distance_performance(self):\n \"\"\"Hamming distance should be fast for many comparisons.\"\"\"\n import time\n\n hash1 = \"a\" * 16\n hash2 = \"b\" * 16\n\n start = time.time()\n for _ in range(10000):\n hamming_distance(hash1, hash2)\n elapsed = time.time() - start\n\n # 10000 comparisons should complete in under 1 second\n self.assertLess(elapsed, 1.0)\n\n def test_find_similar_performance(self):\n \"\"\"find_similar_hashes should be reasonably fast for medium-sized lists.\"\"\"\n import time\n\n target = \"0\" * 16\n # Create a list of 100 hashes\n hash_list = [(f\"img_{i}\", f\"{i:016x}\") for i in range(100)]\n\n start = time.time()\n for _ in range(100):\n find_similar_hashes(target, hash_list, threshold=10)\n elapsed = time.time() - start\n\n # 100 searches over 100 hashes should complete quickly\n self.assertLess(elapsed, 2.0)\n" }, { "path": "api/tests/test_photo_caption_model.py", "content": "from django.test import TestCase\n\nfrom api.models import PhotoCaption\nfrom api.tests.utils import create_test_user, create_test_photo\n\n\nclass PhotoCaptionModelTest(TestCase):\n def setUp(self):\n self.user = create_test_user()\n self.photo = create_test_photo(owner=self.user)\n\n def test_create_photo_caption(self):\n \"\"\"Test creating a PhotoCaption instance\"\"\"\n caption = PhotoCaption.objects.create(\n photo=self.photo, captions_json={\"user_caption\": \"Test caption\"}\n )\n\n self.assertEqual(caption.photo, self.photo)\n self.assertEqual(caption.captions_json[\"user_caption\"], \"Test caption\")\n\n def test_photo_caption_one_to_one_relationship(self):\n \"\"\"Test that PhotoCaption has a one-to-one relationship with Photo\"\"\"\n PhotoCaption.objects.create(\n photo=self.photo, captions_json={\"user_caption\": \"First caption\"}\n )\n\n # Trying to create another caption for the same photo should fail\n with self.assertRaises(Exception):\n PhotoCaption.objects.create(\n photo=self.photo, captions_json={\"user_caption\": \"Second caption\"}\n )\n\n def test_generate_captions_im2txt(self):\n \"\"\"Test generating im2txt captions\"\"\"\n caption = PhotoCaption.objects.create(photo=self.photo)\n\n # This method requires thumbnail access which isn't available in tests\n # We'll test that it returns False when no thumbnail is available\n result = caption.generate_captions_im2txt(commit=False)\n self.assertFalse(result)\n\n def test_save_user_caption(self):\n \"\"\"Test saving user captions\"\"\"\n caption = PhotoCaption.objects.create(photo=self.photo)\n\n # This method requires thumbnail access which isn't available in tests\n # We'll test that it returns False when no thumbnail is available\n result = caption.save_user_caption(\"My beautiful photo\", commit=True)\n self.assertFalse(result)\n\n def test_generate_tag_captions_skips_existing(self):\n \"\"\"Test that generate_tag_captions skips if active model tags already exist\"\"\"\n caption = PhotoCaption.objects.create(photo=self.photo)\n\n # Pre-populate places365 data (the default tagging model)\n caption.captions_json = {\n \"places365\": {\n \"categories\": [\"outdoor\", \"landscape\"],\n \"attributes\": [\"natural\", \"sunny\"],\n \"environment\": \"outdoor\",\n }\n }\n caption.save()\n\n # Should return early since places365 tags already exist\n caption.generate_tag_captions(commit=True)\n caption.refresh_from_db()\n\n self.assertIn(\"places365\", caption.captions_json)\n\n def test_recreate_search_captions_delegates_to_photo_search(self):\n \"\"\"Test that recreate_search_captions delegates to PhotoSearch\"\"\"\n caption = PhotoCaption.objects.create(\n photo=self.photo, captions_json={\"user_caption\": \"Test caption\"}\n )\n\n # This should create a PhotoSearch instance and update search captions\n caption.recreate_search_captions()\n\n # Verify PhotoSearch was created and has search captions\n self.assertTrue(hasattr(self.photo, \"search_instance\"))\n from api.models.photo_search import PhotoSearch\n\n search_instance, created = PhotoSearch.objects.get_or_create(photo=self.photo)\n self.assertIsNotNone(search_instance.search_captions)\n\n def test_captions_json_default_empty_dict(self):\n \"\"\"Test that captions_json defaults to None (nullable field)\"\"\"\n caption = PhotoCaption.objects.create(photo=self.photo)\n\n self.assertIsNone(caption.captions_json)\n\n def test_str_representation(self):\n \"\"\"Test string representation of PhotoCaption\"\"\"\n caption = PhotoCaption.objects.create(\n photo=self.photo, captions_json={\"user_caption\": \"Test\"}\n )\n\n str_repr = str(caption)\n self.assertIn(self.photo.image_hash, str_repr)\n\n def test_cascade_delete_with_photo(self):\n \"\"\"Test that PhotoCaption is deleted when Photo is deleted\"\"\"\n PhotoCaption.objects.create(photo=self.photo)\n photo_id = self.photo.image_hash\n\n self.photo.delete()\n\n with self.assertRaises(PhotoCaption.DoesNotExist):\n PhotoCaption.objects.get(photo_id=photo_id)\n\n def test_multiple_caption_types(self):\n \"\"\"Test storing multiple types of captions\"\"\"\n caption = PhotoCaption.objects.create(\n photo=self.photo,\n captions_json={\n \"user_caption\": \"My photo\",\n \"im2txt\": \"a photo of a landscape\",\n \"places365\": {\n \"categories\": [\"outdoor\"],\n \"attributes\": [\"natural\"],\n \"environment\": \"outdoor\",\n },\n },\n )\n\n self.assertEqual(caption.captions_json[\"user_caption\"], \"My photo\")\n self.assertEqual(caption.captions_json[\"im2txt\"], \"a photo of a landscape\")\n self.assertIn(\"categories\", caption.captions_json[\"places365\"])\n\n def test_update_existing_captions(self):\n \"\"\"Test updating existing captions\"\"\"\n caption = PhotoCaption.objects.create(\n photo=self.photo, captions_json={\"user_caption\": \"Original caption\"}\n )\n\n # Update the caption directly (since save_user_caption requires thumbnails)\n caption.captions_json[\"user_caption\"] = \"Updated caption\"\n caption.save()\n caption.refresh_from_db()\n\n self.assertEqual(caption.captions_json[\"user_caption\"], \"Updated caption\")\n\n def test_empty_captions_json_handling(self):\n \"\"\"Test handling of empty or None captions_json\"\"\"\n caption = PhotoCaption.objects.create(photo=self.photo)\n\n # Should handle empty dict gracefully\n caption.recreate_search_captions()\n\n # Test direct assignment since save_user_caption requires thumbnails\n caption.captions_json = {\"user_caption\": \"\"}\n caption.save()\n self.assertEqual(caption.captions_json[\"user_caption\"], \"\")\n" }, { "path": "api/tests/test_photo_captions.py", "content": "from unittest.mock import patch\n\nfrom django.test import TestCase\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass PhotoCaptionsTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user1 = create_test_user()\n self.user2 = create_test_user()\n self.client.force_authenticate(user=self.user1)\n\n @patch(\n \"api.models.photo_caption.PhotoCaption.generate_captions_im2txt\", autospec=True\n )\n def test_generate_captions_for_my_photo(self, generate_caption_mock):\n generate_caption_mock.return_value = True\n photo = create_test_photo(owner=self.user1)\n\n payload = {\"image_hash\": photo.image_hash}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/generateim2txt/\",\n format=\"json\",\n data=payload,\n headers=headers,\n )\n data = response.json()\n\n self.assertTrue(data[\"status\"])\n\n @patch(\n \"api.models.photo_caption.PhotoCaption.generate_captions_im2txt\", autospec=True\n )\n def test_fail_to_generate_captions_for_my_photo(self, generate_caption_mock):\n generate_caption_mock.return_value = False\n photo = create_test_photo(owner=self.user1)\n\n payload = {\"image_hash\": photo.image_hash}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/generateim2txt/\",\n format=\"json\",\n data=payload,\n headers=headers,\n )\n data = response.json()\n\n self.assertFalse(data[\"status\"])\n\n def test_generate_captions_for_my_photo_of_another_user(self):\n photo = create_test_photo(owner=self.user2)\n\n payload = {\"image_hash\": photo.image_hash}\n headers = {\"Content-Type\": \"application/json\"}\n response = self.client.post(\n \"/api/photosedit/generateim2txt/\",\n format=\"json\",\n data=payload,\n headers=headers,\n )\n data = response.json()\n\n # Returns 404 to avoid leaking existence of other users' photos\n self.assertEqual(404, response.status_code)\n self.assertFalse(data[\"status\"])\n self.assertEqual(\"photo not found\", data[\"message\"])\n" }, { "path": "api/tests/test_photo_lifecycle.py", "content": "\"\"\"\nTests for Photo Lifecycle - deletion, trashing, restoration.\n\nTests verify proper cleanup of:\n- Stack memberships (ManyToMany)\n- Duplicate group memberships (ManyToMany)\n- Empty groups after photo removal\n- Restoration behavior\n\"\"\"\n\nfrom django.test import TestCase\n\nfrom api.models.file import File\nfrom api.models.photo_stack import PhotoStack\nfrom api.models.duplicate import Duplicate\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass PhotoDeletionStackCleanupTestCase(TestCase):\n \"\"\"Tests for photo deletion and stack cleanup.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_manual_delete_clears_stack_membership(self):\n \"\"\"Test that manual_delete removes photo from stacks.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(photo1, photo2, photo3)\n \n self.assertEqual(stack.photos.count(), 3)\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # photo1 should be removed from stack\n photo1.refresh_from_db()\n self.assertEqual(photo1.stacks.count(), 0)\n \n # Stack should still have 2 photos\n stack.refresh_from_db()\n self.assertEqual(stack.photos.count(), 2)\n\n def test_manual_delete_deletes_stack_with_one_remaining(self):\n \"\"\"Test that deleting a photo leaves stack with only 1 photo, stack is deleted.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(photo1, photo2)\n stack_id = stack.id\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Stack should be deleted (only 1 photo remaining)\n self.assertFalse(PhotoStack.objects.filter(id=stack_id).exists())\n \n # photo2 should have no stacks\n photo2.refresh_from_db()\n self.assertEqual(photo2.stacks.count(), 0)\n\n def test_manual_delete_deletes_empty_stack(self):\n \"\"\"Test that deleting all photos in stack deletes the stack.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(photo1, photo2)\n stack_id = stack.id\n \n # Delete both photos\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n photo2.in_trashcan = True\n photo2.save()\n photo2.manual_delete()\n \n # Stack should be deleted\n self.assertFalse(PhotoStack.objects.filter(id=stack_id).exists())\n\n\nclass PhotoDeletionDuplicateCleanupTestCase(TestCase):\n \"\"\"Tests for photo deletion and duplicate group cleanup.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_manual_delete_clears_duplicate_membership(self):\n \"\"\"Test that manual_delete removes photo from duplicate groups.\n \n BUG #12: This test will FAIL if duplicates are not cleared!\n \"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2, photo3)\n \n self.assertEqual(duplicate.photos.count(), 3)\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # photo1 should be removed from duplicate group\n photo1.refresh_from_db()\n self.assertEqual(photo1.duplicates.count(), 0,\n \"Bug #12: manual_delete should clear duplicates\")\n \n # Duplicate group should still have 2 photos\n duplicate.refresh_from_db()\n self.assertEqual(duplicate.photos.count(), 2)\n\n def test_manual_delete_deletes_duplicate_with_one_remaining(self):\n \"\"\"Test that deleting a photo leaves duplicate with only 1 photo, group is deleted.\n \n BUG #12: This test will FAIL if duplicate groups are not cleaned up!\n \"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n duplicate_id = duplicate.id\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Duplicate group should be deleted (only 1 photo remaining)\n self.assertFalse(Duplicate.objects.filter(id=duplicate_id).exists(),\n \"Bug #12: Duplicate group with 1 photo should be deleted\")\n \n # photo2 should have no duplicate groups\n photo2.refresh_from_db()\n self.assertEqual(photo2.duplicates.count(), 0)\n\n def test_manual_delete_deletes_empty_duplicate_group(self):\n \"\"\"Test that deleting all photos in duplicate group deletes the group.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n duplicate_id = duplicate.id\n \n # Delete both photos\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n photo2.in_trashcan = True\n photo2.save()\n photo2.manual_delete()\n \n # Duplicate group should be deleted\n self.assertFalse(Duplicate.objects.filter(id=duplicate_id).exists(),\n \"Bug #12: Empty duplicate group should be deleted\")\n\n\nclass PhotoTrashRestoreTestCase(TestCase):\n \"\"\"Tests for trashing and restoring photos.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_trashed_photo_preserves_stack_membership(self):\n \"\"\"Test that trashing a photo does NOT remove it from stacks.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(photo1, photo2)\n \n # Trash photo1 (not permanent delete)\n photo1.in_trashcan = True\n photo1.save()\n \n # photo1 should still be in stack (just trashed)\n photo1.refresh_from_db()\n self.assertEqual(photo1.stacks.count(), 1)\n \n stack.refresh_from_db()\n self.assertEqual(stack.photos.count(), 2)\n\n def test_trashed_photo_preserves_duplicate_membership(self):\n \"\"\"Test that trashing a photo does NOT remove it from duplicates.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2)\n \n # Trash photo1 (not permanent delete)\n photo1.in_trashcan = True\n photo1.save()\n \n # photo1 should still be in duplicate group (just trashed)\n photo1.refresh_from_db()\n self.assertEqual(photo1.duplicates.count(), 1)\n \n duplicate.refresh_from_db()\n self.assertEqual(duplicate.photos.count(), 2)\n\n def test_restore_photo_from_trash(self):\n \"\"\"Test that restoring a photo from trash works correctly.\"\"\"\n photo1 = create_test_photo(owner=self.user, in_trashcan=True)\n \n # Restore\n photo1.in_trashcan = False\n photo1.save()\n \n photo1.refresh_from_db()\n self.assertFalse(photo1.in_trashcan)\n\n\nclass PhotoInMultipleGroupsTestCase(TestCase):\n \"\"\"Tests for photos that are in multiple stacks and/or duplicate groups.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_photo_in_both_stack_and_duplicate(self):\n \"\"\"Test deleting photo that's in both a stack and duplicate group.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n # Photo1 is in a burst stack with photo2\n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack.photos.add(photo1, photo2)\n \n # Photo1 is also in a duplicate group with photo3\n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo3)\n \n stack_id = stack.id\n duplicate_id = duplicate.id\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n photo1.refresh_from_db()\n \n # Both relationships should be cleared\n self.assertEqual(photo1.stacks.count(), 0)\n self.assertEqual(photo1.duplicates.count(), 0,\n \"Bug #12: Duplicates should be cleared on delete\")\n \n # Both groups should be deleted (only 1 photo remaining in each)\n self.assertFalse(PhotoStack.objects.filter(id=stack_id).exists())\n self.assertFalse(Duplicate.objects.filter(id=duplicate_id).exists(),\n \"Bug #12: Duplicate group should be deleted\")\n\n def test_photo_in_multiple_stacks(self):\n \"\"\"Test photo that's in multiple stacks (different types).\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n # Stack 1: Burst with photo1 and photo2\n stack1 = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n )\n stack1.photos.add(photo1, photo2)\n \n # Stack 2: Manual with photo1 and photo3\n stack2 = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.MANUAL,\n )\n stack2.photos.add(photo1, photo3)\n \n stack1_id = stack1.id\n stack2_id = stack2.id\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Both stacks should be deleted (only 1 photo remaining in each)\n self.assertFalse(PhotoStack.objects.filter(id=stack1_id).exists())\n self.assertFalse(PhotoStack.objects.filter(id=stack2_id).exists())\n\n\nclass DuplicateResolutionCleanupTestCase(TestCase):\n \"\"\"Tests for duplicate resolution affecting photo lifecycle.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_resolve_duplicate_trashes_non_kept_photos(self):\n \"\"\"Test that resolving a duplicate trashes other photos.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n )\n duplicate.photos.add(photo1, photo2, photo3)\n \n # Resolve keeping photo1\n duplicate.resolve(kept_photo=photo1)\n \n # photo2 and photo3 should be trashed\n photo2.refresh_from_db()\n photo3.refresh_from_db()\n self.assertTrue(photo2.in_trashcan)\n self.assertTrue(photo3.in_trashcan)\n \n # photo1 should not be trashed\n photo1.refresh_from_db()\n self.assertFalse(photo1.in_trashcan)\n\n def test_resolve_duplicate_updates_status(self):\n \"\"\"Test that resolving updates duplicate status.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.PENDING,\n )\n duplicate.photos.add(photo1, photo2)\n \n duplicate.resolve(kept_photo=photo1)\n \n duplicate.refresh_from_db()\n self.assertEqual(duplicate.review_status, Duplicate.ReviewStatus.RESOLVED)\n\n\nclass EdgeCasesTestCase(TestCase):\n \"\"\"Edge cases for photo lifecycle.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_delete_photo_not_in_any_group(self):\n \"\"\"Test deleting a photo that's not in any stack or duplicate group.\"\"\"\n photo = create_test_photo(owner=self.user)\n \n photo.in_trashcan = True\n photo.save()\n photo.manual_delete()\n \n photo.refresh_from_db()\n self.assertTrue(photo.removed)\n\n def test_delete_photo_with_no_main_file(self):\n \"\"\"Test deleting a photo without a main_file.\"\"\"\n photo = create_test_photo(owner=self.user)\n photo.main_file = None\n photo.save()\n \n photo.in_trashcan = True\n photo.save()\n \n # Should not crash\n photo.manual_delete()\n \n photo.refresh_from_db()\n self.assertTrue(photo.removed)\n\n def test_stack_primary_photo_deleted(self):\n \"\"\"Test what happens when the primary photo of a stack is deleted.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n photo3 = create_test_photo(owner=self.user)\n \n stack = PhotoStack.objects.create(\n owner=self.user,\n stack_type=PhotoStack.StackType.BURST_SEQUENCE,\n primary_photo=photo1,\n )\n stack.photos.add(photo1, photo2, photo3)\n \n # Delete the primary photo\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Stack should still exist with 2 photos\n stack.refresh_from_db()\n self.assertEqual(stack.photos.count(), 2)\n \n # Primary photo reference might be stale - check it\n # (This tests if there's a bug in primary_photo handling)\n # The primary_photo should ideally be updated or cleared\n\n def test_duplicate_kept_photo_deleted(self):\n \"\"\"Test what happens when the kept_photo of a resolved duplicate is deleted.\"\"\"\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n duplicate = Duplicate.objects.create(\n owner=self.user,\n duplicate_type=Duplicate.DuplicateType.EXACT_COPY,\n review_status=Duplicate.ReviewStatus.RESOLVED,\n kept_photo=photo1,\n )\n duplicate.photos.add(photo1, photo2)\n duplicate_id = duplicate.id\n \n # Now delete the kept photo\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # Duplicate group should be deleted (only 1 photo remaining)\n self.assertFalse(Duplicate.objects.filter(id=duplicate_id).exists(),\n \"Duplicate group with 1 photo should be deleted after kept_photo deletion\")\n\n\nclass SharedFileTestCase(TestCase):\n \"\"\"Tests for photos sharing the same File.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n\n def test_delete_photo_preserves_shared_file(self):\n \"\"\"Test that deleting a photo does not delete a file shared with another photo.\"\"\"\n # Create a shared file\n shared_file = File.objects.create(\n hash=\"shared_file_hash\" + \"a\" * 17,\n path=\"/photos/shared_image.jpg\",\n type=File.IMAGE,\n )\n \n # Create two photos that share the same file\n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n photo1.files.add(shared_file)\n photo1.main_file = shared_file\n photo1.save()\n \n photo2.files.add(shared_file)\n photo2.main_file = shared_file\n photo2.save()\n \n # Verify both photos reference the same file\n self.assertEqual(photo1.main_file.hash, photo2.main_file.hash)\n self.assertEqual(shared_file.photo_set.count(), 2)\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # File should still exist (used by photo2)\n self.assertTrue(File.objects.filter(hash=shared_file.hash).exists(),\n \"File should not be deleted when another photo still uses it\")\n \n # photo2 should still have its main_file\n photo2.refresh_from_db()\n self.assertIsNotNone(photo2.main_file)\n self.assertEqual(photo2.main_file.hash, shared_file.hash)\n \n # photo2's files should still include the shared file\n self.assertTrue(photo2.files.filter(hash=shared_file.hash).exists())\n\n def test_delete_photo_removes_unshared_file(self):\n \"\"\"Test that deleting a photo removes a file only used by that photo.\"\"\"\n import tempfile\n import os\n \n # Create a temp file to simulate a real file\n with tempfile.NamedTemporaryFile(delete=False, suffix='.jpg') as tmp:\n tmp.write(b'test image data')\n temp_path = tmp.name\n \n try:\n # Create a file that's only used by one photo\n unique_file = File.objects.create(\n hash=\"unique_file_hash\" + \"a\" * 18,\n path=temp_path,\n type=File.IMAGE,\n )\n \n photo = create_test_photo(owner=self.user)\n photo.files.add(unique_file)\n photo.main_file = unique_file\n photo.save()\n \n # Verify only one photo uses this file\n self.assertEqual(unique_file.photo_set.count(), 1)\n \n # Delete the photo\n photo.in_trashcan = True\n photo.save()\n photo.manual_delete()\n \n # File should be deleted from database\n self.assertFalse(File.objects.filter(hash=unique_file.hash).exists(),\n \"File should be deleted when no other photos use it\")\n \n # Physical file should be deleted\n self.assertFalse(os.path.exists(temp_path),\n \"Physical file should be removed from disk\")\n finally:\n # Cleanup in case test fails\n if os.path.exists(temp_path):\n os.remove(temp_path)\n\n def test_delete_photo_with_shared_main_file_different_from_files(self):\n \"\"\"Test deleting when main_file is shared but files M2M has unique files.\"\"\"\n # Shared main_file\n shared_main = File.objects.create(\n hash=\"shared_main_hash\" + \"a\" * 18,\n path=\"/photos/main.jpg\",\n type=File.IMAGE,\n )\n \n # Unique sidecar file for photo1 only\n unique_sidecar = File.objects.create(\n hash=\"unique_sidecar_hash\" + \"a\" * 15,\n path=\"/photos/sidecar.xmp\",\n type=File.METADATA_FILE,\n )\n \n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n # Both photos share main_file\n photo1.main_file = shared_main\n photo1.files.add(shared_main, unique_sidecar)\n photo1.save()\n \n photo2.main_file = shared_main\n photo2.files.add(shared_main)\n photo2.save()\n \n # Delete photo1\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n # shared_main should still exist (used by photo2)\n self.assertTrue(File.objects.filter(hash=shared_main.hash).exists())\n \n # unique_sidecar should be deleted (only used by photo1)\n self.assertFalse(File.objects.filter(hash=unique_sidecar.hash).exists())\n \n # photo2 should still reference shared_main\n photo2.refresh_from_db()\n self.assertEqual(photo2.main_file.hash, shared_main.hash)\n\n def test_delete_last_photo_using_shared_file(self):\n \"\"\"Test that file is deleted when the last photo using it is deleted.\"\"\"\n shared_file = File.objects.create(\n hash=\"eventually_orphan\" + \"a\" * 17,\n path=\"/photos/shared.jpg\",\n type=File.IMAGE,\n )\n \n photo1 = create_test_photo(owner=self.user)\n photo2 = create_test_photo(owner=self.user)\n \n photo1.files.add(shared_file)\n photo1.main_file = shared_file\n photo1.save()\n \n photo2.files.add(shared_file)\n photo2.main_file = shared_file\n photo2.save()\n \n # Delete photo1 - file should remain\n photo1.in_trashcan = True\n photo1.save()\n photo1.manual_delete()\n \n self.assertTrue(File.objects.filter(hash=shared_file.hash).exists())\n \n # Delete photo2 - file should now be deleted (no photos using it)\n photo2.in_trashcan = True\n photo2.save()\n photo2.manual_delete()\n \n # File should be deleted from database (no physical file to check)\n self.assertFalse(File.objects.filter(hash=shared_file.hash).exists(),\n \"File should be deleted when the last photo using it is deleted\")\n" }, { "path": "api/tests/test_photo_list_without_timestamp.py", "content": "from django.test import TestCase\nfrom django.utils import timezone\nfrom rest_framework.test import APIClient\n\nfrom api.tests.utils import create_test_photos, create_test_user\n\n\nclass PhotoListWithoutTimestampTest(TestCase):\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n\n def test_retrieve_photos_without_exif_timestamp(self):\n now = timezone.now()\n create_test_photos(number_of_photos=1, owner=self.user, added_on=now)\n create_test_photos(\n number_of_photos=1, owner=self.user, added_on=now, exif_timestamp=now\n )\n\n response = self.client.get(\"/api/photos/notimestamp/\")\n json = response.json()\n\n self.assertEqual(response.status_code, 200)\n self.assertEqual(1, len(json[\"results\"]))\n" }, { "path": "api/tests/test_photo_metadata.py", "content": "\"\"\"\nComprehensive tests for PhotoMetadata model and API.\n\nTests cover:\n- PhotoMetadata model fields and properties\n- MetadataFile model\n- MetadataEdit model for change tracking\n- API endpoints (retrieve, update, history, revert)\n- Bulk metadata operations\n- Edge cases and error handling\n\"\"\"\n\nimport uuid\nfrom unittest.mock import patch\n\nfrom django.test import TestCase\nfrom rest_framework import status\nfrom rest_framework.test import APIClient\n\nfrom api.models.photo_metadata import MetadataEdit, MetadataFile, PhotoMetadata\nfrom api.tests.utils import create_test_photo, create_test_user\n\n\nclass PhotoMetadataModelTestCase(TestCase):\n \"\"\"Tests for PhotoMetadata model functionality.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.photo = create_test_photo(owner=self.user)\n\n def test_create_metadata_basic(self):\n \"\"\"Test creating basic PhotoMetadata.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n aperture=2.8,\n iso=100,\n focal_length=50.0,\n camera_model=\"Canon EOS R5\",\n )\n \n self.assertEqual(metadata.photo, self.photo)\n self.assertEqual(metadata.aperture, 2.8)\n self.assertEqual(metadata.iso, 100)\n self.assertEqual(metadata.focal_length, 50.0)\n self.assertEqual(metadata.camera_model, \"Canon EOS R5\")\n\n def test_metadata_source_choices(self):\n \"\"\"Test metadata source choices.\"\"\"\n self.assertEqual(PhotoMetadata.Source.EMBEDDED, \"embedded\")\n self.assertEqual(PhotoMetadata.Source.SIDECAR, \"sidecar\")\n self.assertEqual(PhotoMetadata.Source.USER_EDIT, \"user_edit\")\n self.assertEqual(PhotoMetadata.Source.COMPUTED, \"computed\")\n\n def test_resolution_property(self):\n \"\"\"Test resolution property.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n width=1920,\n height=1080,\n )\n \n self.assertEqual(metadata.resolution, \"1920x1080\")\n\n def test_resolution_property_missing_dimensions(self):\n \"\"\"Test resolution property with missing dimensions.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n width=1920,\n height=None,\n )\n \n self.assertIsNone(metadata.resolution)\n\n def test_megapixels_property(self):\n \"\"\"Test megapixels calculation.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n width=8256,\n height=5504,\n )\n \n # 8256 * 5504 = 45,441,024 pixels ≈ 45.4 MP\n self.assertEqual(metadata.megapixels, 45.4)\n\n def test_megapixels_property_missing_dimensions(self):\n \"\"\"Test megapixels with missing dimensions.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n width=None,\n height=None,\n )\n \n self.assertIsNone(metadata.megapixels)\n\n def test_has_location_property_with_gps(self):\n \"\"\"Test has_location with GPS data.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n gps_latitude=40.7128,\n gps_longitude=-74.0060,\n )\n \n self.assertTrue(metadata.has_location)\n\n def test_has_location_property_without_gps(self):\n \"\"\"Test has_location without GPS data.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n )\n \n self.assertFalse(metadata.has_location)\n\n def test_has_location_partial_gps(self):\n \"\"\"Test has_location with only latitude.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n gps_latitude=40.7128,\n gps_longitude=None,\n )\n \n self.assertFalse(metadata.has_location)\n\n def test_camera_display_make_and_model(self):\n \"\"\"Test camera_display with both make and model.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n camera_make=\"Canon\",\n camera_model=\"EOS R5\",\n )\n \n self.assertEqual(metadata.camera_display, \"Canon EOS R5\")\n\n def test_camera_display_model_includes_make(self):\n \"\"\"Test camera_display when model already includes make.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n camera_make=\"Canon\",\n camera_model=\"Canon EOS R5\",\n )\n \n # Should not duplicate make\n self.assertEqual(metadata.camera_display, \"Canon EOS R5\")\n\n def test_camera_display_only_model(self):\n \"\"\"Test camera_display with only model.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n camera_model=\"EOS R5\",\n )\n \n self.assertEqual(metadata.camera_display, \"EOS R5\")\n\n def test_lens_display(self):\n \"\"\"Test lens_display property.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n lens_make=\"Canon\",\n lens_model=\"RF 50mm F1.2L\",\n )\n \n self.assertEqual(metadata.lens_display, \"Canon RF 50mm F1.2L\")\n\n def test_lens_display_model_includes_make(self):\n \"\"\"Test lens_display when model includes make.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n lens_make=\"Canon\",\n lens_model=\"Canon RF 50mm F1.2L\",\n )\n \n self.assertEqual(metadata.lens_display, \"Canon RF 50mm F1.2L\")\n\n def test_version_increments(self):\n \"\"\"Test version field increments on save.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n aperture=2.8,\n )\n \n self.assertEqual(metadata.version, 1)\n \n metadata.aperture = 4.0\n metadata.version += 1\n metadata.save()\n \n metadata.refresh_from_db()\n self.assertEqual(metadata.version, 2)\n\n def test_raw_data_json_fields(self):\n \"\"\"Test raw EXIF/XMP/IPTC JSON fields.\"\"\"\n raw_data = {\n \"EXIF:Make\": \"Canon\",\n \"EXIF:Model\": \"EOS R5\",\n \"EXIF:ISO\": 100,\n }\n \n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n raw_exif=raw_data,\n )\n \n self.assertEqual(metadata.raw_exif, raw_data)\n\n def test_keywords_json_field(self):\n \"\"\"Test keywords JSON field stores list.\"\"\"\n keywords = [\"landscape\", \"sunset\", \"nature\"]\n \n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n keywords=keywords,\n )\n \n metadata.refresh_from_db()\n self.assertEqual(metadata.keywords, keywords)\n\n\nclass MetadataFileModelTestCase(TestCase):\n \"\"\"Tests for MetadataFile model.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.photo = create_test_photo(owner=self.user)\n\n def test_file_type_choices(self):\n \"\"\"Test file type choices exist.\"\"\"\n self.assertEqual(MetadataFile.FileType.XMP, \"xmp\")\n self.assertEqual(MetadataFile.FileType.JSON, \"json\")\n self.assertEqual(MetadataFile.FileType.EXIF, \"exif\")\n self.assertEqual(MetadataFile.FileType.OTHER, \"other\")\n\n def test_source_choices(self):\n \"\"\"Test source choices exist.\"\"\"\n self.assertEqual(MetadataFile.Source.ORIGINAL, \"original\")\n self.assertEqual(MetadataFile.Source.SOFTWARE, \"software\")\n self.assertEqual(MetadataFile.Source.LIBREPHOTOS, \"librephotos\")\n self.assertEqual(MetadataFile.Source.USER, \"user\")\n\n\nclass MetadataEditModelTestCase(TestCase):\n \"\"\"Tests for MetadataEdit model.\"\"\"\n\n def setUp(self):\n self.user = create_test_user()\n self.photo = create_test_photo(owner=self.user)\n\n def test_create_edit_record(self):\n \"\"\"Test creating a metadata edit record.\"\"\"\n edit = MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=\"Old Title\",\n new_value=\"New Title\",\n )\n \n self.assertEqual(edit.photo, self.photo)\n self.assertEqual(edit.user, self.user)\n self.assertEqual(edit.field_name, \"title\")\n self.assertEqual(edit.old_value, \"Old Title\")\n self.assertEqual(edit.new_value, \"New Title\")\n self.assertFalse(edit.synced_to_file)\n\n def test_edit_records_ordered_by_created_at(self):\n \"\"\"Test edit records are ordered by creation time.\"\"\"\n edit1 = MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=None,\n new_value=\"First\",\n )\n edit2 = MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=\"First\",\n new_value=\"Second\",\n )\n \n edits = list(MetadataEdit.objects.filter(photo=self.photo))\n # Most recent first\n self.assertEqual(edits[0].id, edit2.id)\n self.assertEqual(edits[1].id, edit1.id)\n\n\nclass PhotoMetadataAPITestCase(TestCase):\n \"\"\"Tests for PhotoMetadata API endpoints.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.other_user = create_test_user()\n self.client.force_authenticate(user=self.user)\n \n self.photo = create_test_photo(owner=self.user)\n self.metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n aperture=2.8,\n iso=100,\n focal_length=50.0,\n camera_make=\"Canon\",\n camera_model=\"EOS R5\",\n width=8256,\n height=5504,\n )\n\n def test_retrieve_metadata(self):\n \"\"\"Test retrieving metadata for a photo.\"\"\"\n response = self.client.get(f\"/api/photos/{self.photo.id}/metadata\")\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"aperture\"], 2.8)\n self.assertEqual(data[\"iso\"], 100)\n self.assertEqual(data[\"camera_model\"], \"EOS R5\")\n\n def test_retrieve_metadata_by_image_hash(self):\n \"\"\"Test retrieving metadata using image_hash.\"\"\"\n response = self.client.get(f\"/api/photos/{self.photo.image_hash}/metadata\")\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n\n def test_retrieve_metadata_creates_if_missing(self):\n \"\"\"Test retrieve creates metadata if it doesn't exist.\"\"\"\n photo2 = create_test_photo(owner=self.user)\n \n # Delete any auto-created metadata\n PhotoMetadata.objects.filter(photo=photo2).delete()\n \n response = self.client.get(f\"/api/photos/{photo2.id}/metadata\")\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n # Metadata should now exist\n self.assertTrue(PhotoMetadata.objects.filter(photo=photo2).exists())\n\n def test_retrieve_metadata_other_user_forbidden(self):\n \"\"\"Test retrieving other user's photo metadata is forbidden.\"\"\"\n other_photo = create_test_photo(owner=self.other_user)\n \n response = self.client.get(f\"/api/photos/{other_photo.id}/metadata\")\n \n self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)\n\n def test_update_metadata(self):\n \"\"\"Test updating metadata fields.\"\"\"\n response = self.client.patch(\n f\"/api/photos/{self.photo.id}/metadata\",\n {\"title\": \"My Beautiful Photo\", \"rating\": 5},\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n \n self.metadata.refresh_from_db()\n self.assertEqual(self.metadata.title, \"My Beautiful Photo\")\n self.assertEqual(self.metadata.rating, 5)\n\n def test_update_creates_edit_history(self):\n \"\"\"Test updating metadata creates edit history.\"\"\"\n response = self.client.patch(\n f\"/api/photos/{self.photo.id}/metadata\",\n {\"caption\": \"A stunning sunset\"},\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n \n # Check edit history was created\n edits = MetadataEdit.objects.filter(photo=self.photo)\n self.assertTrue(edits.exists())\n edit = edits.first()\n self.assertEqual(edit.field_name, \"caption\")\n self.assertEqual(edit.new_value, \"A stunning sunset\")\n\n def test_get_edit_history(self):\n \"\"\"Test getting edit history.\"\"\"\n # Create some edits\n MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=None,\n new_value=\"First\",\n )\n MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=\"First\",\n new_value=\"Second\",\n )\n \n response = self.client.get(f\"/api/photos/{self.photo.id}/metadata/history\")\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertIn(\"results\", data)\n self.assertEqual(data[\"count\"], 2)\n self.assertEqual(len(data[\"results\"]), 2)\n\n def test_revert_edit(self):\n \"\"\"Test reverting a specific edit.\"\"\"\n # Set initial value\n self.metadata.title = \"Original Title\"\n self.metadata.save()\n \n # Create an edit\n edit = MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=\"Original Title\",\n new_value=\"Changed Title\",\n )\n self.metadata.title = \"Changed Title\"\n self.metadata.save()\n \n # Revert the edit\n response = self.client.post(f\"/api/photos/{self.photo.id}/metadata/revert/{edit.id}\")\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n \n self.metadata.refresh_from_db()\n self.assertEqual(self.metadata.title, \"Original Title\")\n\n def test_revert_creates_new_edit_record(self):\n \"\"\"Test reverting creates a new edit record.\"\"\"\n edit = MetadataEdit.objects.create(\n photo=self.photo,\n user=self.user,\n field_name=\"title\",\n old_value=\"Original\",\n new_value=\"Changed\",\n )\n self.metadata.title = \"Changed\"\n self.metadata.save()\n \n initial_count = MetadataEdit.objects.filter(photo=self.photo).count()\n \n self.client.post(f\"/api/photos/{self.photo.id}/metadata/revert/{edit.id}\")\n \n final_count = MetadataEdit.objects.filter(photo=self.photo).count()\n self.assertEqual(final_count, initial_count + 1)\n\n def test_revert_nonexistent_edit(self):\n \"\"\"Test reverting nonexistent edit returns 404.\"\"\"\n fake_edit_id = uuid.uuid4()\n \n response = self.client.post(f\"/api/photos/{self.photo.id}/metadata/revert/{fake_edit_id}\")\n \n self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)\n\n def test_unauthenticated_request(self):\n \"\"\"Test unauthenticated requests return 401.\"\"\"\n self.client.force_authenticate(user=None)\n \n response = self.client.get(f\"/api/photos/{self.photo.id}/metadata\")\n \n self.assertEqual(response.status_code, status.HTTP_401_UNAUTHORIZED)\n\n\nclass BulkMetadataAPITestCase(TestCase):\n \"\"\"Tests for bulk metadata operations.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n \n self.photo1 = create_test_photo(owner=self.user)\n self.photo2 = create_test_photo(owner=self.user)\n self.photo3 = create_test_photo(owner=self.user)\n \n self.meta1 = PhotoMetadata.objects.create(\n photo=self.photo1,\n camera_model=\"Canon R5\",\n rating=3,\n )\n self.meta2 = PhotoMetadata.objects.create(\n photo=self.photo2,\n camera_model=\"Nikon Z9\",\n rating=4,\n )\n\n def test_bulk_get_metadata(self):\n \"\"\"Test getting metadata for multiple photos.\"\"\"\n photo_ids = f\"{self.photo1.id},{self.photo2.id}\"\n \n response = self.client.get(f\"/api/photos/metadata/bulk?photo_ids={photo_ids}\")\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertIn(str(self.photo1.id), data)\n self.assertIn(str(self.photo2.id), data)\n\n def test_bulk_get_no_photo_ids(self):\n \"\"\"Test bulk get without photo_ids returns error.\"\"\"\n response = self.client.get(\"/api/photos/metadata/bulk\")\n \n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_bulk_get_max_100_photos(self):\n \"\"\"Test bulk get with >100 photos returns error.\"\"\"\n # Create 101 fake photo IDs\n photo_ids = \",\".join([str(uuid.uuid4()) for _ in range(101)])\n \n response = self.client.get(f\"/api/photos/metadata/bulk?photo_ids={photo_ids}\")\n \n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n self.assertIn(\"Maximum 100\", response.json()[\"error\"])\n\n def test_bulk_update_metadata(self):\n \"\"\"Test bulk updating metadata.\"\"\"\n response = self.client.patch(\n \"/api/photos/metadata/bulk\",\n {\n \"photo_ids\": [str(self.photo1.id), str(self.photo2.id)],\n \"updates\": {\"rating\": 5},\n },\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"updated_count\"], 2)\n \n self.meta1.refresh_from_db()\n self.meta2.refresh_from_db()\n self.assertEqual(self.meta1.rating, 5)\n self.assertEqual(self.meta2.rating, 5)\n\n def test_bulk_update_no_photo_ids(self):\n \"\"\"Test bulk update without photo_ids returns error.\"\"\"\n response = self.client.patch(\n \"/api/photos/metadata/bulk\",\n {\"updates\": {\"rating\": 5}},\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_bulk_update_no_updates(self):\n \"\"\"Test bulk update without updates returns error.\"\"\"\n response = self.client.patch(\n \"/api/photos/metadata/bulk\",\n {\"photo_ids\": [str(self.photo1.id)]},\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n\n def test_bulk_update_invalid_fields(self):\n \"\"\"Test bulk update with invalid fields returns error.\"\"\"\n response = self.client.patch(\n \"/api/photos/metadata/bulk\",\n {\n \"photo_ids\": [str(self.photo1.id)],\n \"updates\": {\"iso\": 100}, # ISO is not allowed for bulk edit\n },\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)\n self.assertIn(\"Invalid fields\", response.json()[\"error\"])\n\n def test_bulk_update_creates_edit_history(self):\n \"\"\"Test bulk update creates edit history for each photo.\"\"\"\n initial_count = MetadataEdit.objects.count()\n \n self.client.patch(\n \"/api/photos/metadata/bulk\",\n {\n \"photo_ids\": [str(self.photo1.id), str(self.photo2.id)],\n \"updates\": {\"title\": \"Bulk Title\"},\n },\n format=\"json\",\n )\n \n # Should have 2 new edit records (one per photo)\n self.assertEqual(MetadataEdit.objects.count(), initial_count + 2)\n\n def test_bulk_update_other_user_photos_ignored(self):\n \"\"\"Test bulk update ignores other user's photos.\"\"\"\n other_user = create_test_user()\n other_photo = create_test_photo(owner=other_user)\n \n response = self.client.patch(\n \"/api/photos/metadata/bulk\",\n {\n \"photo_ids\": [str(self.photo1.id), str(other_photo.id)],\n \"updates\": {\"rating\": 5},\n },\n format=\"json\",\n )\n \n self.assertEqual(response.status_code, status.HTTP_200_OK)\n data = response.json()\n self.assertEqual(data[\"updated_count\"], 1) # Only our photo\n\n\nclass PhotoMetadataEdgeCasesTestCase(TestCase):\n \"\"\"Edge case tests for PhotoMetadata.\"\"\"\n\n def setUp(self):\n self.client = APIClient()\n self.user = create_test_user()\n self.client.force_authenticate(user=self.user)\n self.photo = create_test_photo(owner=self.user)\n\n def test_metadata_with_special_characters(self):\n \"\"\"Test metadata fields handle special characters.\"\"\"\n metadata = PhotoMetadata.objects.create(\n photo=self.photo,\n title=\"Photo with émojis 📷 and ünïcödé\",\n caption=\"\",\n )\n \n metadata.refresh_from_db()\n self.assertIn(\"émojis\", metadata.title)\n self.assertIn(\"📷\", metadata.title)\n self.assertIn(\"